Video games often represent the frontier of any new technology. Many of the most popular applications in the initial iPhone app store were games. Today’s virtual reality devices are dominated by video games.
Artificial intelligence seems poised to upend the video game business and entertainment more broadly.
On the third episode of our six-part Cerebral Valley podcast series, Max Child, James Wilsterman, and I game out how artificial intelligence could reshape the media we consume.
It helps that Max and James are the co-founders of Volley, which builds AI-enabled games. They develop many of the most popular voice games on the Amazon Alexa and smart TV platforms like Roku.
Max and James have been deep in the trenches of conversational-style gaming and have spent a lot of time thinking about how humans interact with ever smarter computers.
In the second half of the episode, I talk with Menlo Ventures partner Amy Wu, who focuses on gaming and consumer investments, and Keith Kawahata, a former executive at Wargaming, AppLovin, and Kabam, who now has a stealth artificial intelligence gaming startup.
Wu helps to articulate a three-part thesis on how artificial intelligence might change the gaming business. (1) artificial intelligence will help with the creation of the game art and graphics, (2) AI can create more sophisticated non-player characters, and (3) AI can help write the code of the game itself.
One of the things that I was struck by from the conversation is that games may have a big leg up in implementing artificial intelligence over movies — because people interact so much more with a gaming, giving it tons of data to react to. While TikToks can learn what small populations of people like and what an individual likes over a long time, a game could learn a lot about a user in a single play session.
Of course, there are real hurdles left standing. Most notably, text-to-image generation so far is mostly two-dimensional. Despite everything that’s happened, image generation models aren’t just whipping out 3-D levels that are ready to play.
And it could be a while until non-player characters are as smart as humans. But imagine playing a game of Grand Theft Auto or Red Dead Redemption and the characters roaming around the game are self-aware agents with their own thoughts and drives.
Give it a listen
P.S. I’m on my honeymoon right now in Japan. I was working frantically to record these episodes before I left. My chief of staff Riley Konsella is sending the episodes out for me while I’m gone. If you need anything while I’m away, you should email Riley.
Thanks in advance for being understanding that this newsletter is slowing down for my honeymoon. I’m going to dedicate myself to relaxing over the next two weeks so that I come back hungrier than ever.
Get full access to Newcomer at www.newcomer.co/subscribe
00:00:10
Hey, welcome back. This is Cerebral Valley.
00:00:13
I'm Eric Newcomer. My guest on this episode will be
00:00:16
Amy Wu of Menlo Ventures, who was an investor at FTX and deep
00:00:21
in the crypto world and is now hardcore into AI Gaming
00:00:25
Investments, and Keith Kawahata, who's the CEO of a stealth AI
00:00:30
gaming startup and has been a mobile gaming exec at a bunch of
00:00:34
different. Companies.
00:00:35
So we need to talk to him. I need to hear more about that.
00:00:37
Yeah, exactly. I have Max Child and James
00:00:40
Wilsterman, my Cerebral Valley Co hosts and Volley Co founders
00:00:45
here. This episode is very much in
00:00:48
their domain. We're going to talk about
00:00:50
artificial intelligence and its impact on entertainment, and
00:00:54
some of the time, particularly their field gaming.
00:00:57
Welcome back. Happy to be here.
00:00:59
Explain Volley and the how sort of the AI games world as you
00:01:05
guys experience it. Sure.
00:01:08
Well, we're definitely in a very specific corner of AI games, and
00:01:13
we're built around voice control.
00:01:14
So we build games that you talk to rather than play with a
00:01:18
controller or, you know, using a touch screen like your
00:01:21
smartphone. So we have games that you can
00:01:23
play on. You know, an AI device like an
00:01:25
Amazon Alexa as well as that's where the majority of them are.
00:01:28
That is where the majority of our users are today.
00:01:30
But actually the fastest growing part of our company is on Smart
00:01:32
TV's, Roku and Fire TV. And we have different types of
00:01:36
games you can play with your voice.
00:01:37
We have game shows like Jeopardy, so we have like
00:01:40
Jeopardy on Roku and you get a big board in front of you like
00:01:43
you're actually on Jeopardy and you can buzz in with your remote
00:01:46
and you know say give me US Capitals for 200, Alex.
00:01:49
And you know what is Sacramento and?
00:01:52
Yeah, we have a number of other kind of voice control gaming
00:01:54
experiences, including storytelling, kind of fiction,
00:01:57
Choose your own adventure that you control with your voice, as
00:01:59
well as some music stuff. Yes, Sir.
00:02:02
Is that still? Are people still into that or is
00:02:04
it? Yeah, I mean, it's actually
00:02:06
pretty popular. That's our Choose Your own
00:02:07
adventure fiction game to it gives you like two options and
00:02:11
you go one way or the other. It's been out for six years.
00:02:14
It's still going, but it's not. I would say it's not the cutting
00:02:17
edge of all. Yeah, stop talking about.
00:02:18
That but no. But I think it's an interesting
00:02:20
tie to where we are today with AI because the reason we built
00:02:24
yes sire at the time was to really simplify what you could
00:02:28
say to the device, really dumb it down a lot so that it never
00:02:32
misunderstood you or you know, couldn't process your language
00:02:36
input. So it was 2 choices, you could
00:02:38
either say yes or no and that would affect the game.
00:02:41
I think where we are today with AII think it's much, you know,
00:02:44
the technology has improved drastically where it, you know,
00:02:47
it recognizes things better. But also now with LLMS, we can
00:02:51
take in a lot wider sort of inputs around natural language.
00:02:57
So you know, obviously part of the volley bet is which has a
00:03:02
huge AI component because there's sort of the voice
00:03:04
recognition and all that and we could probably do a whole
00:03:07
episode on and you guys are obviously big advocates of voice
00:03:10
being sort of one of the main ways we'll talk to AIS.
00:03:14
But I'm curious like when you think of gaming and AI
00:03:17
generally, I mean I think of like creation or what are the
00:03:20
pieces where you actually think artificial intelligence and
00:03:23
generative AII mean assets? You know, like what are the
00:03:27
parts of games you see AI fitting into?
00:03:30
I think there's a lot of different layers of the stack as
00:03:32
we would call it, right? And I think that, you know, our
00:03:35
company's built on some stuff that really started working.
00:03:38
Five or six years ago in the sort of last generation of you
00:03:41
know, deep learning, machine learning, AI, whatever you want
00:03:43
to call it at the time. And that was just speech
00:03:45
recognition getting pretty good all of a sudden and you know
00:03:48
Alexa being one of the big steps forward there.
00:03:50
And then secondly like what we call like natural language
00:03:52
processing, which is like turning human phrasing into
00:03:56
meaning, right. And both of those got pretty
00:03:58
good like 7-8 ten years ago, however you want to measure it
00:04:01
and that was sort of what enabled our company to exist was
00:04:04
like. You could build a game on an
00:04:06
Alexa device and when you said you know something to a a
00:04:10
storytelling game guest sire, it would understand it and we could
00:04:12
do the right thing, right. But to your point, what's
00:04:15
exciting about, you know, the last year and a half and the
00:04:18
reason we have a cerebral valley conference is the sort of the
00:04:21
generative element of AI has really taken this hugely
00:04:24
forward. The idea that we can create new
00:04:26
things that that you know, AI can create things that.
00:04:29
Humans either couldn't create as quickly or they can do remixes
00:04:32
that have never been done before.
00:04:33
And there's a few different things that AI can generate.
00:04:36
One is just text, right? It can generate scripts for a
00:04:40
storytelling game or a movie, right?
00:04:42
It can generate images like you were talking about assets within
00:04:46
a game, so like art that you could use within a game, you
00:04:49
could have an AI generate it, like a stable diffusion or a mid
00:04:52
journey. And then one thing that's
00:04:54
relevant for us is also it can generate.
00:04:56
Fake voices, right? It can create synthesized voices
00:04:59
very easily. You know 10 years ago.
00:05:01
Does Amazon let you do that right now?
00:05:03
Yeah. So Alexa was one of the first,
00:05:04
you know, high quality synthesized voices.
00:05:07
But ten years ago it cost like a quarter $1 to make a
00:05:11
good synthesized voice. And then today with the sort of
00:05:14
generative transformer wave, it costs 5 bucks now.
00:05:17
So it's come down by, I think with the script we're going to,
00:05:20
you know, you know, if I'm too busy, they're going to slice in.
00:05:23
Yeah, right. Scott's going to fill in my
00:05:25
voice as needed. So yeah, so that's what I
00:05:28
actually said on this. Yeah, exactly right.
00:05:31
You know, some people call it like deep fake voices where I,
00:05:33
you know, I think it's like synthesize voices is kind of an
00:05:35
amazing. I mean, it's literally like
00:05:37
100 times better than it was 10 years ago, which is like
00:05:40
really big. Yeah, one of our most.
00:05:42
It's recent games on Roku where you can play Jeopardy.
00:05:45
It's actually hosted by one of our employees.
00:05:48
We synthesize his voice and it sounds amazing.
00:05:51
It sounds really realistic, but we were able to do that because
00:05:53
of the advancement in the technology.
00:05:56
Right. And so in in all kinds of
00:05:58
gaming, whether or not you believe voice is the input or
00:06:00
controllers of the input or it's mobile gaming or whatever.
00:06:03
I think the output stuff, the content of the game, the art and
00:06:07
the assets and the voices and the way you hear the experience
00:06:11
and the music as well we haven't talked about as well.
00:06:13
Like you could synthesize music and games.
00:06:15
Those are all like completely going to change or changing as
00:06:19
we speak across the whole world of gaming from like AAA, you
00:06:23
know, crazy console games all the way down to like more casual
00:06:26
experiences and and more sort of family game night type stuff,
00:06:29
which is a lot of what we built. James, which one do you think
00:06:32
will be the actual most revolutionary?
00:06:34
Or, you know, there's a lot there, but which piece of it are
00:06:38
you most excited about? So Max and I really like this
00:06:42
scene in her, which people probably don't remember very
00:06:46
well, but it is a scene in HER where the main character is
00:06:51
playing a video game and it's really interesting because the
00:06:57
character in the game is like a foul mouthed alien.
00:07:01
Sort of child. And the player has to kind of
00:07:06
interact with it, with natural language and convince it to help
00:07:08
him on a quest. But you know, it gets the
00:07:13
character is not really helpful. It's like very adversarial and I
00:07:19
think that opens up some new kind of ideas and gaming and an
00:07:23
AI. Right now, every assistant is
00:07:26
trying to be extremely helpful. Whereas once you kind of put
00:07:29
this in an entertainment context, you can have
00:07:31
adversarial AI that are, you know, leading you astray.
00:07:35
What what do you think feels like I, you know, we see I
00:07:39
haven't played them, but you know, you hear about the No
00:07:41
Man's Sky, like they're all these space games.
00:07:43
I don't know if they're not using LLMS necessarily, but
00:07:45
they're trying to generate sort of.
00:07:47
Audit generation. Yeah, right.
00:07:49
There's a lot like, let's throw technology at this stuff at, you
00:07:53
know, human storytelling is still, you know, the best thing
00:07:56
out there, and a lot of this stuff can end up feeling
00:08:00
gimmicky. What do you which feels close
00:08:03
that that would actually be useful?
00:08:05
I think there's, I think there's three layers of like complexity
00:08:08
of the content generation you can do in gaming and in sort of
00:08:11
entertainment more broadly. I think the first layer is you
00:08:13
can just generate some content. Let's say it's, you know, a
00:08:17
script in a game or in a movie, or in our case like a trivia
00:08:22
question, let's say. Right?
00:08:24
And. You can generate it from an LLM.
00:08:26
You can ask an LLM for a Jeopardy question right now.
00:08:29
You'll get something pretty good.
00:08:31
You know, we would look it over and have a human in the loop
00:08:34
doing some editing, but then we we put it into one of our trivia
00:08:38
games, right? And so that would be like, hey,
00:08:40
the humans who work for Volley or whatever game company now
00:08:43
have to do a lot a little better.
00:08:44
Yeah, they just got a little better, right?
00:08:46
They're they're faster writers. Or let's say they're an artist,
00:08:48
right? And they can generate artwork
00:08:50
way faster, or generate a bunch of ideas really quickly and then
00:08:54
discard some of them. And iterate on some of them, but
00:08:56
in the end there's a review process and then these things
00:08:58
ship into the games, right? That I think is like happening
00:09:02
as we speak, right? I mean within our company, like
00:09:04
that's happening as we speak. Within other gaming companies.
00:09:07
I'm aware of people already doing generative art for in this
00:09:11
kind of pipeline where they use a mid journey or stable
00:09:14
diffusion or whatever or some specialized tool and then they
00:09:16
review it and then they ship it out to to their players, right?
00:09:19
So. That's like Level 1 and that's
00:09:21
happening like for sure. But it is basically like
00:09:24
augmenting the human in the loop, basically.
00:09:26
Then there's like the second layer, which we talked a little
00:09:29
bit on the first episode about, but it's like you start
00:09:32
personalizing the content within a gaming experience in real time
00:09:35
or you were talking about a TikTok feed being personalized
00:09:38
for each person. But a gaming content is often
00:09:40
easier to sort of think about where it's OK, let's use the
00:09:42
Jeopardy example again. Let's say that.
00:09:44
Every question within the Jeopardy game that I'm playing
00:09:46
is is literally decided for me by a large language model idiot
00:09:51
and he needs to win enough for him to get claims.
00:09:54
Well, not just difficulty adjustments though.
00:09:57
But what if you were able to say, you know, if you're really
00:10:00
like obsessed with Harry Potter, you could just make the whole
00:10:03
board about Harry Potter. It would generate instantly,
00:10:05
right? Or if you're studying for.
00:10:08
You know an AP physics class? Like, why not put that into our
00:10:12
game and play a Jeopardy board of AP physics classes?
00:10:15
Right. I think those will be shipping
00:10:16
out more and more in the next three, 612 months where you're
00:10:19
creating personalized experiences and or you're
00:10:22
enabling user generated kind of content experiences in games.
00:10:25
And then the last level was sort of what James was talking about
00:10:28
where AI are characters in games, right.
00:10:31
Let's just say you're playing Zelda and you run into some
00:10:33
dwarf or whatever and like. It just like talks to you right?
00:10:37
And you're talking to the dwarf and like you it has some mission
00:10:41
in mind which is to give you a potion or whatever and but you
00:10:44
can ask you like anything. You could be like hey like where
00:10:46
should I go next? Like why is my sword like you
00:10:48
know wearing down like how many you know coins is it going to
00:10:52
cost me to upgrade my shield or whatever right?
00:10:54
You can ask me questions but in the end there's some object of
00:10:56
that conversation right? And I think non player
00:10:59
characters like that. Generated by large language
00:11:02
models is, you know, it's really expensive is honestly the
00:11:05
biggest thing holding it back right now.
00:11:07
If you want to run a really good large language model like at GBT
00:11:10
at at that kind of scale, having millions of gamers playing it,
00:11:13
it's like really expensive. But I think that in a couple
00:11:17
years or a year or two or whatever, there will be ways to
00:11:20
sort of minify these models or focus these models or just the
00:11:24
hardware will get cheaper or whatever, and then you'll start
00:11:25
to see the stuff in. World of Warcraft or Zelda or
00:11:28
whatever, you know, and and just to zoom out, like why are we
00:11:31
talking about gaming besides obviously you guys being
00:11:33
experts. I do.
00:11:34
I I mean gaming in so much of technology is sort of the
00:11:38
bleeding edge of what consumers experience.
00:11:40
Like obviously in AI right now there's a ton of infrastructure,
00:11:44
wonky stuff that the regular person's never going to see.
00:11:46
But I do think gaming and entertainment will be where a
00:11:51
lot of this stuff emerges. You know, we we've talked about
00:11:54
before, you know sort of the chat apps which are sort of.
00:11:57
Interactive entertainment, Yeah, right.
00:11:59
Well, I mean, NVIDIA, which is the biggest company in AI right
00:12:01
now, literally spent decades selling GPUs to gaming
00:12:05
companies, right? I mean, NVIDIA exists because of
00:12:07
gaming, right? I mean, to your point, gaming is
00:12:09
always it's, you know, it's the bleeding edge of entertainment
00:12:12
because it is fundamentally just like interactive entertainment.
00:12:15
And you know, when you add technology to entertainment, you
00:12:18
get gaming basically, right. So I think that.
00:12:20
Yeah, I and I also just think that if you look at industry
00:12:23
dynamics, gaming now makes more money than movies and TV and
00:12:28
music like all put together. It's it is the biggest
00:12:30
entertainment industry in the world by a huge margin.
00:12:33
So it is like fundamentally how people actually really
00:12:36
experience entertainment these days.
00:12:38
I wanted to pull be the journalist and annoy my friends.
00:12:42
Would you use at Volley, mid Journey or Dolly assets or what?
00:12:47
Yeah. What are your lawyers telling
00:12:48
you in terms of like actually putting these things in the
00:12:52
game? Because some of this is going to
00:12:53
be like, which big company? You're not as big, but which big
00:12:56
company is going to take the risk on this sort of thing,
00:12:58
right? Yeah, I think that's a really
00:13:00
interesting question and I honestly I think we are
00:13:03
discussing this a lot internally to give a real answer.
00:13:06
It's it is a topic of debate inside our company to say the
00:13:09
least. And I think that you know, I
00:13:11
think the maybe we'll get to the whole copyright discussion later
00:13:13
in in the the podcast, but I think that the question of
00:13:17
whether or not. Training a huge image AI on a
00:13:21
bunch of like images you scraped off websites is legal or not is
00:13:25
like an open question. And so I think we would try to
00:13:30
use something where we had some good information about the
00:13:34
provenance of the training data. And I don't know, James, you've
00:13:37
obviously been thinking about this even more than I have.
00:13:39
Yeah. What?
00:13:39
What do you think? Yeah, I think that right now the
00:13:42
answer is no. We don't use AI generated images
00:13:45
in our products and I think we are internally.
00:13:49
Debating that and trying to understand where this image data
00:13:52
came from and how do we feel about the legal implications of
00:13:56
that. I think that there's so Dolly
00:13:58
Three was announced recently. I think they specifically
00:14:02
mentioned there would be like robust opt out features for
00:14:05
artists and their training data. And I think they will start to
00:14:09
see more and more, you know, like Max said, tools that
00:14:13
actually have to delineate the provenance of the data and then
00:14:16
maybe that opens up. You know a legal path for it,
00:14:20
but it's really an open question.
00:14:22
Did you guys see the Getty Images just launched a image
00:14:25
generation tool? Yeah.
00:14:26
So that's pretty exciting. I think the progress in 100%
00:14:29
licensed image generation is very fast, right.
00:14:33
And so I think if you are concerned about that side of
00:14:36
things, I think that you now have Adobe Firefly, which is all
00:14:39
licensed. You have the Getty generation
00:14:42
tool, which is all licensed. So I guess I just sort of think
00:14:44
this isn't. There's going to be that much of
00:14:46
a problem even in a year from now, because you're going to
00:14:48
have pretty robust, you know, license generative tools.
00:14:52
I do think like an interesting subsequent question here though
00:14:56
is these are these types of questions are being litigated in
00:14:59
the courts, right? I mean, there are a bunch of
00:15:01
lawsuits occurring. Sarah Silverman and people.
00:15:04
Sarah Silverman and other artists, and I don't know, What
00:15:07
do you guys think? Do we think that there's a
00:15:09
clear? Outcome of that based on current
00:15:12
law or do you think they're you know that these it seems very
00:15:16
serious like it I I do think there's got to be some copyright
00:15:19
I mean not being a lawyer but just.
00:15:22
Do you, do you think that applies to LLMS to then?
00:15:24
I mean, that's essentially what Sarah Silverman is saying,
00:15:27
right? Yeah, Yeah.
00:15:28
Yeah, I hope. I mean.
00:15:30
Would you have an issue if I could create a newcomer article
00:15:32
based on scraping all the newcomer articles?
00:15:34
I mean, you're a journalist, so a lot of your original content
00:15:37
is from reporting, not from your writing.
00:15:39
But yeah, I mean definitely, yeah.
00:15:42
So you you do have it. I mean that's part of, I mean so
00:15:45
yeah, this is exactly I the writers strike, we want to talk
00:15:48
about it, let's talk about it because it's directly relevant
00:15:51
to this question. I think you know even me sort of
00:15:55
being very bullish on AI and like around it, you can sort of
00:15:58
laugh a little bit at the writers making like prioritizing
00:16:02
AI sort of like a top priority. It feels paranoia.
00:16:05
On the other hand, all the things we've said in this
00:16:08
conversation, it feels very current that, you know, they
00:16:10
basically won protections that you know, you know, chat product
00:16:14
products can't replace their work.
00:16:16
Like the writers get to choose if they use it and sort of a
00:16:19
bunch of different. And they got that basically.
00:16:22
Yeah, that was the resolution, right.
00:16:23
I mean, do you think they needed it?
00:16:24
I don't know. I guess that was that a real and
00:16:26
present danger. So I have a good story on this
00:16:29
front, which is I was meeting a big like Hollywood studio honcho
00:16:33
like six months ago and. He was like, oh, you know, what
00:16:36
do you think about all this AI stuff?
00:16:38
And I was like, oh, I think it's pretty cool.
00:16:39
You know, you prayed on with chat, JPEG, whatever.
00:16:41
And he's like, yeah, like, I asked it to come up with some,
00:16:44
like, new movie ideas for our studio.
00:16:46
And he's came out with seven ideas.
00:16:47
And he's honestly like five of them were pretty good.
00:16:50
He was like, he was like, I think I might use like five of
00:16:52
them, honestly. And just like a real Hollywood
00:16:55
studio guy. Yeah, exactly.
00:16:57
I never know what good ideas of you in the face.
00:17:01
I mean, I'm just saying, I was in the room where the guy was
00:17:03
like, Oh yeah, like these are. These ideas were pretty good my
00:17:06
friend and and it didn't produce screenplays for him for the
00:17:11
record, but it had some concepts that were going to work for him,
00:17:14
so. Yeah, well, I think the
00:17:16
agreement that was struck here, you know, a lot of it had to do
00:17:19
with can you take, can the studios take these old scripts
00:17:22
that they own that the writers wrote but, you know, gave away
00:17:26
those copyrights essentially to the studios.
00:17:29
And they use them to train future scripts.
00:17:31
And I think what is, you know, going kind of unsaid there is
00:17:36
that, you know, everyone believes that in the current
00:17:40
legal system, legal frameworks like the studios have every
00:17:43
right to go use those scripts for training and write future
00:17:46
scripts with. However, you know, the unions
00:17:48
were able to negotiate sort of like.
00:17:51
Retroactive rights for their writers on scripts that they
00:17:55
already sold to this they make a broad point on this that is
00:17:58
eating up at me. I mean, first of all, it seems
00:18:00
right now like foundation models are becoming pretty
00:18:03
commoditized. There are several of them.
00:18:05
Like maybe open AI stays far enough ahead and finds other
00:18:09
modes, but like in some ways like the creative expression is
00:18:15
more unique, right. Lord of the Rings is like more
00:18:18
singular. And yeah, I just don't think
00:18:22
like somebody should be able to just steal.
00:18:24
That was a great creative work. And just because people didn't
00:18:27
foresee this technology being around at the time, like, why
00:18:31
should just like these technologists who found who are
00:18:34
the first ones to figure out how to steal the idea in a way they
00:18:36
could get away with, be the ones to profit off of it.
00:18:40
I'm on the creative side here. I don't think that.
00:18:44
You know, I said Volley could create a game based on Middle
00:18:48
Earth, and that's just because of existing IP protections that
00:18:52
prevent us from doing that whether or not we write the
00:18:54
content with AI or not, right? And likewise, nobody's can go
00:18:58
and write a whole, you know, a bunch of newcomer articles and
00:19:02
claim they're part of newcomer because you have you own the IP.
00:19:06
So the question is more like can I, you know, write a brand new
00:19:10
IP script that you know has some?
00:19:13
Training data of the original Lord of the Rings series.
00:19:16
And and it gets to the one of the core idea is an LLM.
00:19:19
Because obviously, like a human being, you wouldn't say you
00:19:21
can't go and read Lord of the Rings and then go write another
00:19:25
book having like been changed, right.
00:19:26
It's like a point that's basically what all art is, you
00:19:29
know? But with the machine you're
00:19:30
like, well, you basically just pull little bits and pieces from
00:19:34
everything and that's stealing, right.
00:19:36
So yeah, I mean, they're big, like philosophical, like how,
00:19:39
you know, how different is it from how a human being sort of
00:19:42
operates? Yeah, I mean I think you know we
00:19:45
one thing we think about on copyright side, but also I think
00:19:48
on the sort of like conceptual side of you know stealing versus
00:19:52
remixing versus original content is like.
00:19:55
Like in the US, we actually have this pretty robust framework on
00:19:57
copyright. We call it, you know the the
00:19:58
fair use test of whether or not you can reuse some portion of
00:20:01
copyrighted work, right? And when you can, it's called
00:20:05
fair use, and when you can't, you're infringing and and you
00:20:07
can be sued or whatever, right? And there's a bunch of key
00:20:10
elements of the fair use test that I actually think are all
00:20:12
really interesting and relevant in the AI discussion and the LLM
00:20:15
discussion, And the image and music generation is also really
00:20:19
hot topics here, like the key pieces are like.
00:20:23
How transformative is the use of the original work?
00:20:26
You know, is the new work compared to the original work?
00:20:28
So how much have you changed? Basically, like how different it
00:20:30
is actually fairly important. How much of the original work
00:20:33
you are using or replicating is also fairly important.
00:20:37
Whether or not the use is commercial is also like a big
00:20:39
part. Are you selling this thing for
00:20:41
money or is it just like a fun project or is it educational?
00:20:45
And then the last piece which we've all kind of been talking
00:20:47
about is are you impinging upon the market for the original
00:20:53
work, which is like really important, right.
00:20:56
So yeah, so to the Lord of the Rings discussion, like if I
00:20:59
created a, you know, a fantasy novel that said in a sort of,
00:21:03
you know, vaguely related universe of the Lord of the
00:21:06
Rings, but I didn't call it the Lord of the Rings.
00:21:08
And, you know, I wasn't stealing any of the characters.
00:21:11
And, you know, Jr. Tolkien's estate wasn't going to
00:21:14
sell less Lord of the Rings books because I came out with
00:21:16
any fantasy novel then that's a pretty good use.
00:21:19
You know, example, like fair use for.
00:21:20
I'm not really impinging upon the original.
00:21:22
And it's also quite transformative, right?
00:21:24
I think music is the one where this is like the most stark in
00:21:27
some cases because there's these AI tik tokers that are making
00:21:31
like fake Drake music, right. And like fake like Taylor Swift
00:21:35
music and stuff, right. And the music industry obviously
00:21:38
doesn't like this. They're having them trying to
00:21:39
have them all taken down, right. And it's an interesting
00:21:42
question. If I make a fake Drake song,
00:21:45
like, obviously that's not very transformative, right?
00:21:49
But am I impinging upon the market for Drake, right?
00:21:52
Am I like, am I capturing some portion of Drake love in the
00:21:56
universe, you know, with my fake Drake song?
00:21:59
And I kind of think so, yeah, I sort of agree.
00:22:01
Like, I I think that the fake Drake tracks are probably
00:22:03
copyright violations, right? But then the second question is
00:22:06
like, what if I make a completely new piece of music,
00:22:09
right, that is trained on every pop music, you know, piece of
00:22:13
pop music ever, but doesn't sound like any existing artist
00:22:16
in a meaningful way, right? I'm not pretending it's Taylor
00:22:19
Swift, I'm not pretending it's straight, whatever.
00:22:21
It's just a new band that you know, it's the Max James and
00:22:24
Eric Band that puts out awesome pop music because it's a really
00:22:27
cool AI creation based on existing pop music.
00:22:30
Is that fair use? If you train it on all the pop
00:22:32
music, I'm sure this will be litigated quite heavily, but I
00:22:36
think you can make an argument that it's fair use.
00:22:37
You know there's also a. Question that AI is new and
00:22:40
there should be new regulations. I I'm very torn on what
00:22:44
independent law what ought to be the case.
00:22:46
On the one hand it's it would be nice to have a system where
00:22:49
artists got paid proportionally to how much their work you know
00:22:53
sort of created the circumstances for the new AI
00:22:56
based invention. On the other hand, just, you
00:22:58
know, all those sort of barriers, you know?
00:23:01
Just create lots of hurdles to new ideas and you know, it's
00:23:04
like the same type of thinking that prevents like, you know,
00:23:07
remixes and sampling and all that stuff when that's what the
00:23:10
consumer ultimately wants and enjoys.
00:23:13
And also in the context of an LM or a transformer model like it
00:23:17
seems pretty hard to. Tie this new output of the song
00:23:20
like that. Yeah, attribute it back to the
00:23:23
thousand artists, right, who had some factor who deserve some
00:23:27
fraction of a penny, right. I just think it's not practical,
00:23:34
but if but if I make fake Van Gogh art on mid journey, should
00:23:37
the Van Gogh estate get a cut of?
00:23:39
It feels like in some just world.
00:23:41
Any, any. I mean, yeah, maybe.
00:23:42
It's just interesting, right? You can do it.
00:23:44
I mean, so, but yeah, Van Gogh, if it's like Calvin and Hobbes.
00:23:48
Yeah, definitely. You know, Yeah.
00:23:50
I mean, if you call it Calvin and Hobbes for.
00:23:51
Sure, but I think this is just we don't even need to know
00:23:55
whether it was Air I or not. We already have these laws in
00:23:57
place. If IA Court looks at it and it
00:23:59
looks like Calvin and Hobbes or it is called.
00:24:02
You're already in trouble. Like, why does it matter if it
00:24:05
was generated by AI? SO that's kind of why I push
00:24:07
back against the idea that we do need new laws, because I've
00:24:10
never heard a better framework than this one.
00:24:12
I guess I'm not saying that don't exist, but at least this
00:24:14
one covers like a lot of the relevant considerations here,
00:24:17
right? And the big one being, are you
00:24:20
impacting the original art? You know the value of the
00:24:23
original work in a negative way, right?
00:24:25
I think that's super important. And it's hard to figure out,
00:24:28
even in an AI world, unless you go whole hog and you're just
00:24:31
like, yeah, scraping any copyrighted content is illegal
00:24:33
for an AI. It's OK Well, yeah, that's a new
00:24:35
law. We.
00:24:36
Can make right? Isn't that you know?
00:24:38
There's certain humans you know with the fan fiction have, you
00:24:40
know, dump, you know, played around with stuff.
00:24:42
But I mean there, it's not commercial, right?
00:24:45
It's not commercial. It's not commercial.
00:24:46
Yeah, right. Let's say we agree that this
00:24:49
there don't need to be many new laws here and kind of the
00:24:51
existing fair use framework is sufficient I guess.
00:24:55
That does seem to lead to a lot of potential challenges in, you
00:25:00
know, the creative fields and the workforce that generates art
00:25:04
and for games and scripts, for films and, you know, songs going
00:25:09
forward, right? So do we think that's
00:25:11
inevitable, essentially that the like creative industries are
00:25:15
going to need less people to generate this type of content?
00:25:21
I mean, it's hard to say. I mean, part of what Max was
00:25:23
saying in the beginning is sort of the democratizing effect that
00:25:26
it could allow. Or sorry, Jamie, you know, you
00:25:29
guys run together. No.
00:25:32
Yeah. Yeah, exactly.
00:25:33
It's giving, sort of. People with less capabilities,
00:25:37
the tools to to do more, which could just just big companies,
00:25:41
right? I I guess I just think there is
00:25:42
a different discussion to be had around people who own copyright
00:25:46
and how that affects the the, you know, future works versus
00:25:48
people who don't own copyright but who work for companies
00:25:51
generating creative output, you know, and how that what their
00:25:56
lives will look like going forward, what their work streams
00:25:58
will look like. I mean, we don't want a world
00:26:02
where all future creative content is just echoes of what
00:26:07
came before and humans stop. But that that also kind of gets
00:26:11
to whether LLMS can be creative. If you believe that LLMS are
00:26:15
capable of original output, then that won't be the case, right?
00:26:19
Yeah, you're right. There's an argument that we
00:26:21
could be in a renaissance of quickly all these new art types
00:26:24
we've never experienced, It doesn't feel like.
00:26:26
Like Dolly. And stuff are producing that
00:26:28
yet, right, do you think? This, I mean, I don't know.
00:26:31
I think that if you prompt these LMS correctly and spend some
00:26:35
time on it, it's not like 100% of the time you'll get creative
00:26:38
output, but I believe it is possible to get it to generate
00:26:42
new ideas. And you want to be in a museum.
00:26:45
You can put up a green painting on the wall and call it art, but
00:26:50
you need to have a really pretentious explanation for what
00:26:52
the point was and where It's in conversation with everything
00:26:55
else and these machines cannot produce.
00:26:57
Any sort of like narrative for why they're doing what they're
00:27:00
doing. So it is not.
00:27:01
It's not art. This is here the original.
00:27:05
Green painting man. But Max, to Max's point, if the
00:27:08
studio exec can just use this tool to generate commercially
00:27:14
viable ideas for movies that are better than anyone on the staff
00:27:18
has generated, I I think that. Alone is an interesting.
00:27:22
Right. And commercial generation is oh,
00:27:24
we should do Star Wars plus. But that's what that person is
00:27:27
optimizing for. Theoretically, I mean, but yeah.
00:27:30
All right. I wanted to ask if you could
00:27:34
like you know dream up, you know using generative AI some like
00:27:41
sequel to a favorite work. Like what would you like to see
00:27:44
or what would most excite you that feels like in reach because
00:27:49
of? Some of the stuff we've talked
00:27:51
about. I guess I can answer that.
00:28:01
I feel like, you know TV shows that maybe ended for their prime
00:28:07
or something, right? I'm trying to think what a good
00:28:09
example of that is. But or.
00:28:11
Here's a name. Firefly is a famous.
00:28:13
Yeah, Firefly, everyone says. I liked it.
00:28:15
It wasn't life changing. But let's talk about lost or
00:28:17
something, right? If I could, I thought the ending
00:28:20
was not great, but not as terrible as the Internet
00:28:24
believes. But I would be happy to go rerun
00:28:26
the last season, right? Wouldn't that be cool to see if
00:28:28
you could create? You know 100 different versions
00:28:31
of the ending of Lost. I love that idea.
00:28:33
I love the idea that, you know, like, first of all, I quit Lost
00:28:37
at the end of Season 2 being like they can never pay this
00:28:39
off. So I'm very proud of that.
00:28:41
But the idea that like you know with Game of Thrones and stuff
00:28:43
where like clearly the author really can like puzzle out the
00:28:48
like solution that they need, but it feels like there should
00:28:51
be some right answer that you like.
00:28:53
Throw an LLM at it. Maybe there's a right answer
00:28:55
that exists. I think if they could redo the
00:28:57
last season in Game of Thrones, that would not be a terrible
00:28:59
idea. Whether or not, yeah, like
00:29:02
whether or not the writer, whether or not RR Martin is
00:29:04
giving them guidance or someone else.
00:29:06
But I think that would be a good one to rerun.
00:29:09
Use the generative AI as the excuse.
00:29:11
Who cares if it's actually better than, like some random
00:29:14
writer? It's Oh well, we have this new
00:29:15
technique. I mean, it's not really a sequel
00:29:17
for me, but I think it would be possible to create just like
00:29:21
really awesome music that I love, right?
00:29:23
That's personalized, right? I think that again, I'm not like
00:29:27
if it's trained on a bunch of copyrighted material and artist
00:29:29
should be paid for it. And I'm cool with that.
00:29:30
I'm, I'm figuring out how to pay this AI company.
00:29:33
But just like the pure product itself, it would be really cool
00:29:36
to have personalized music, you know, on demand that really just
00:29:40
hit the sweet spot for me or whatever, right?
00:29:42
I mean, I just think that would be so cool.
00:29:45
And I think not that hard from a, I mean conceptual
00:29:49
perspective, just train an LL or not an LLM.
00:29:52
But as music generation model, everything Max loves, everything
00:29:55
he's listened to more than 10 times in his life or whatever in
00:29:57
his Spotify library, right? I just think that would be
00:29:59
awesome. I don't know.
00:30:00
That'd be really cool. Well.
00:30:01
You could also imagine a world. I mean, you're talking, I think
00:30:04
about. Novel sounding art, right?
00:30:07
Not the Drake example in that. Yeah, right.
00:30:09
I don't need another Kanye album.
00:30:11
We kind of work out that well. But another Beatles album?
00:30:15
I think a lot of people there would be a market for that,
00:30:17
right? But yeah, I do think maybe this
00:30:20
ends up looking a lot like the fan fiction community is where.
00:30:23
Anyone can create a new Beatles album.
00:30:25
They just can't sell it, right? You can't.
00:30:27
I mean, that would be pretty sweet too.
00:30:29
I know, right? Shareware Beatles albums like
00:30:31
you just have a ranking of the best ones basically, I mean.
00:30:35
Yeah, Do you forecasting cultural, do you think?
00:30:38
I mean, you can imagine, yeah, people get to like refresh you,
00:30:42
you know, you got a personalized version of every episode or you
00:30:44
don't like the take. It's like a Magic 8 ball and
00:30:47
you're like, Oh my God, take it again.
00:30:49
Like I that seems so culturally, like, hollow to me.
00:30:52
I guess it's very exciting to me where you could create new art.
00:30:56
Like, I think my answer to this would just be, I mean, I just
00:30:58
think the NPC like creative NPCS, like where you're actually
00:31:02
engaging with them like any game, just like any open world,
00:31:05
you know, you can just imagine playing.
00:31:07
I don't know. Any Horizon 0 Dawn where like
00:31:10
they're real characters and they have sort of.
00:31:13
Their own agendas that are different every game.
00:31:15
And to bring this back to the last fake tech trend, which was
00:31:18
like the Metaverse, right, The Metaverse populated by a bunch
00:31:21
of characters that you know weren't that flexible or
00:31:24
interesting, I don't think is that exciting.
00:31:26
But the Metaverse in a world where you can run into an
00:31:28
infinite number of digital characters that have interesting
00:31:31
stories that you can go on interesting quests and missions
00:31:34
with and and compete and collaborate with.
00:31:36
I just think it's. I mean I want to hang out with
00:31:37
other people in a in an ARVR world as well.
00:31:40
But I think hanging out with really cool, interesting non
00:31:43
player characters that are powered by LLM is just a really
00:31:46
fun idea too. And so I just think Ready Player
00:31:49
1:00 is way more fun if half the people in Ready Player One are
00:31:52
like interesting, conversational, you know, AI
00:31:54
driven characters. Also because AI will be much
00:31:57
happier to spend tons of time in the Metaverse than regular
00:32:01
people, and so you could actually populate it with them
00:32:03
versus forcing strap on. There you go.
00:32:06
They're always on. We haven't talked about movies
00:32:09
as much as I thought we would. I'm curious, I guess I'm
00:32:13
personally skeptical like the movie business is anywhere is
00:32:16
near affected in the near term like.
00:32:19
Maybe sort of I what they sort of you know development has
00:32:23
changed but it feels like. I do think like just that seems
00:32:27
to be just driven it seemed, yeah, I mean the generative
00:32:32
video space is like very early right from the current
00:32:35
technology, but it's still amazing and so they.
00:32:38
My prediction on the first episode was that Tik Tok's being
00:32:42
revolutionary. So yeah, right.
00:32:44
I mean I just don't like, I think this sort of bums me out
00:32:47
as someone who watched a lot of movies growing up.
00:32:49
But I just sort of think the movie business is in like a
00:32:51
secular decline. And I think all these other
00:32:53
types of content are like eating it up, whether it's like short
00:32:57
form video or television series or TikTok or AI generated
00:33:02
TikTok. And I mean as we discussed at
00:33:03
the beginning, gaming is bigger than all these things put
00:33:05
together because it turns out like having some interactivity
00:33:08
and some form of, you know, play as part of your entertainment
00:33:11
experience makes it more fun and makes it more engaging.
00:33:13
So I just think that movies are a sort of almost like
00:33:17
anachronistic type of content. I know that's a really hot take.
00:33:19
But this idea that you have a 2 hour thing that like you don't
00:33:23
interact with, that has if a sequel comes out, if ever, it
00:33:27
comes out five years later and it's also only two hours long
00:33:30
and it's not personalized in any way.
00:33:32
And also the business model is falling apart because nobody
00:33:35
wants to go to theaters anymore. So they can only make Marvel
00:33:37
movies basically now. So it's just not, It's caught
00:33:41
between a lot of factors that are bad for the future of what
00:33:45
we think of as films or movies, right.
00:33:47
And I think, you know, TV has already been eating movies
00:33:50
throughout our lifetime where I think most of the like prestige
00:33:54
dramas that have come out in the last, you know, 10 or 20 years
00:33:56
of a lot of a moon on TV, which like never happened before 2000.
00:34:00
And then I think also I think gaming will start to eat into
00:34:03
that. And I think, you know, short
00:34:04
form video, whether it's TikTok or YouTube or or some AI
00:34:08
generated version of these things, we'll just start to eat
00:34:09
into that also. So I guess, yeah, the sort of
00:34:11
depressing take is I don't think movies like matter anywhere near
00:34:14
as much as they once did, which is why I think the other stuff's
00:34:16
more interesting to talk about. Yeah, that's some very
00:34:20
depressing on movies. I'm sorry.
00:34:22
It feels like areas where you can iterate.
00:34:25
I mean, short form video is just so perfect because you can
00:34:28
really test it out like the machines can learn from it.
00:34:31
Whereas yeah, movies require a ton of intention and sort of
00:34:35
broad thinking, which is still the domain of humans.
00:34:38
So I don't know if everyone in the movie business would be so
00:34:40
depressed by what you're saying. It's like, Oh yeah, the real art
00:34:44
sort of the long form thinking is a little more insulated than
00:34:48
the let's run an AB test and see how people react to things.
00:34:52
And so there there's some good in that too, definitely.
00:34:56
I mean, I still think there will be great movies throughout our
00:34:57
lifetime, but I think that where the new tools are coming into
00:35:01
play is in categories where, yeah, maybe the movie makers
00:35:04
aren't that interested in. Or if we really believe in the
00:35:07
extension of the gaming discussion we had at the
00:35:09
beginning and we can create extremely intelligent NPCS.
00:35:15
Can you just put them in simulations that are fun to
00:35:18
watch in the same way people watch reality shows or, you
00:35:22
know, watch films and TV's? The stories are so compelling
00:35:26
that but they're just occurring because there's a simulation
00:35:29
running. I don't know.
00:35:30
Maybe that's where. We are.
00:35:30
That would be a fun The Real Housewives of Open AI here.
00:35:33
Yeah, exactly. Patent that, yeah.
00:35:37
Yeah, that sounds good. Yeah, little Big Brother, but
00:35:40
just with AI running around. Yeah.
00:35:42
Or. And then and then, you know,
00:35:44
you're essentially able to iterate through thousands of
00:35:47
different reality scenarios, right?
00:35:49
Throw an island in there, or a rose ceremony, whatever it is.
00:35:54
All right. Stay tuned.
00:35:55
I have Amy and Keith up next. Keith, Amy, welcome to the show.
00:36:04
Thanks so much for joining me. Thanks so much, Eric.
00:36:07
Yeah. Thanks for having us.
00:36:08
Amy, I I thought, I mean, we'd start off with you just as.
00:36:12
Sort of the venture capitalist who sees things broadly, can you
00:36:17
talk a bit about your interest in gaming?
00:36:22
Yeah, it's it's a it's a lifelong passion really.
00:36:26
I haven't worked in the game industry but have been, you
00:36:28
know, playing games. I was a gamer growing up and
00:36:31
play a lot of, you know, early, like, you know, first person
00:36:35
shooters like the Doom and the Wolfenstein.
00:36:39
Era games to like CS and then Diablo.
00:36:42
Era to more of the modern day and and as an investor you know.
00:36:47
What an industry. I love that you get.
00:36:48
It's like, oh, we did all these fun things and that's part of
00:36:51
the credit of it. Like, I should spend more time
00:36:53
in gaming. Anyway, yeah, it's so true.
00:36:56
It really is. I mean it's pretty important I
00:36:58
think because especially if you decide to invest in content,
00:37:02
first of all it's really difficult and and there are
00:37:05
people play testing really like hundreds if not thousands of
00:37:07
games every year. But it is really important to to
00:37:11
play them because then you can get kind of a feel for what
00:37:14
really exceptional looks like in different genres.
00:37:17
Yeah, in in terms on the investing site, didn't you, you
00:37:19
invested in Epic at one point, is that right?
00:37:22
That's right. Yeah.
00:37:23
Well, do the first foray was actually when I was at Insight.
00:37:27
This is like 10 years ago now and Insight was majority owner
00:37:31
of Jag X Runescape and and that was like an early experience
00:37:36
because I I still remember when. When there was like a big
00:37:40
monetization change in the game and everyone sort of the players
00:37:43
were like outraged and and they kind of saw that you know
00:37:47
Insight was a big shareholder and so then you know there were
00:37:51
like bomb threats called into the fund and yeah people were
00:37:55
player. I mean, yeah, it was a great
00:37:59
example of just how intensely passionate gamers are about
00:38:03
their games and and changes to it like.
00:38:06
And so then later on you know I joined Livespeed and one of my
00:38:09
mandates was actually investing games and games there.
00:38:12
And we did early stage investments in companies like
00:38:14
you know Faraway which is web three side really experienced
00:38:18
game makers from Scopely and then and then also like AAA
00:38:21
studios like 1047 and and lastly games and others and also growth
00:38:26
stage there was Epic Games and then also triple dot studios
00:38:29
such as mobile Studio. And then on on the crypto side,
00:38:32
I mean, I'm sure there. Other things, but I mean it's
00:38:35
you Go Labs, right, the board ape company, I mean part of the
00:38:38
investment thesis. There.
00:38:40
What is it you got right. OK yeah that you were gonna that
00:38:45
they were are we're going to build out sort of games and sort
00:38:50
of in an entertainment world around to that is that right?
00:38:52
Or how much was games part of investing in them?
00:38:56
In the you go round and this was early last year I think that.
00:39:01
First and foremost, the exciting thing was that with board of
00:39:04
Yacht Club and then you know they then acquired Crypto Punks
00:39:08
and a couple of other brands. It was the I would say the blue
00:39:12
chip brand in the NFT space. And so there was a lot of value
00:39:16
in that and that they could turn into potential like consumer
00:39:20
products that a lot that both NFT holders and aspirational NFT
00:39:24
holders could use and love. And and so games was a a big
00:39:28
part of that strategy all. Right, Keith.
00:39:31
Tell, tell. Just give us a little bit of
00:39:33
your overview. You have a stealth start up,
00:39:34
which you know, amazing. You know when I'm prying that
00:39:39
you're going to be able ducking and dodging on some of that, but
00:39:42
give give people. A little bit of your background
00:39:44
in the games industry. Yeah, sure.
00:39:48
So actually just before, before I got into games, I did five
00:39:53
years in the US Army, artillery and and Special Operations and I
00:39:58
I bring that up because it it's. It's unexpectedly like kind of
00:40:01
come back as as part of my life in game.
00:40:04
But for the last 1215 years I've been in the games industry at
00:40:10
startups, mostly later stage startups.
00:40:13
So you know, past the kind of 0 to one point more at the growth
00:40:16
phase as well as like larger conglomerates.
00:40:19
So companies like Kabam in the early days of web, Facebook and
00:40:23
then mobile gaming, Gree International, the subsidiary
00:40:26
of, you know, the large social network Gree in Japan,
00:40:30
Wargaming. I'm actually visiting some old
00:40:32
friends and colleagues here at the Wargaming office right now,
00:40:36
makers of, you know, the phenomenal World of Tanks.
00:40:39
And then most recently was Ads app Loving where I led the games
00:40:43
business there. To our speaking the language of
00:40:46
actual in particular games, because I think you know not
00:40:49
everybody follows like the games business.
00:40:51
What were the games that ended up defining a lot of your career
00:40:54
in terms of spending time? Yeah, so you know I've, I love
00:41:00
every and I I mean this very generally.
00:41:02
I love every game that I've been a part of in in one way or
00:41:05
another. Even those that have like if he
00:41:07
was kind of sharing resulted in some funny real world
00:41:10
circumstances like getting getting offices firebombed and
00:41:14
threatened and stuff like that. Even that is actually just a a
00:41:16
symptom of the passion that exists in the industry.
00:41:18
And I I share that passion as a gamer as well.
00:41:20
Maybe not to the violent ends, but yeah, so.
00:41:23
So games. Games is what people would know.
00:41:25
So marquee games for me, you know, the first game I ever
00:41:29
worked on was a game called Last Chaos, which is a Korean
00:41:33
developed MMRPG that we were publishing over in the West.
00:41:37
No, not many people know about it.
00:41:39
Like if you weren't one of the 100 or so people that tried
00:41:41
it, like you wouldn't know. But it's it.
00:41:43
It speaks to the longevity and some other elements of the
00:41:46
industry because that game is still alive today, believe it or
00:41:49
not. But game the people would know.
00:41:52
So King is a Camelot would have been you know an early mobile
00:41:56
actually first Facebook and then early mobile 4X strategy game
00:42:02
that I spent some time on and then also at Kabayam, you know
00:42:06
as a part of a number of IP tie in games as well.
00:42:10
So we did Marvel, contents of Champions.
00:42:13
We did a couple of while I was there.
00:42:15
We did a couple of Fast and the Furious titles as well.
00:42:20
And then let's see next Wargaming.
00:42:24
So World of Tanks, World of Warships.
00:42:26
You know, there's pretty big titles in the shooter space or
00:42:29
in the strategic shooter spaces, as I used to say back then.
00:42:33
And then most recently, oh sorry, I almost forgot.
00:42:35
Actually, the most played game that I've ever had the pleasure
00:42:39
to work on and and with is Subway Surfers with the.
00:42:42
Oh my God. Every TikTok video right is
00:42:45
exactly. I mean it is the most
00:42:47
downloaded. I believe it is the most single
00:42:49
downloaded game ever. Wow.
00:42:51
It, you know, it counts its installs in the multiple
00:42:54
billiards and what's your portion of the human race like
00:42:57
has played that game at one time or another?
00:42:59
Do you have a typical role in these games or like how would
00:43:01
you describe what you tend, what the role you?
00:43:04
Yeah, so early, early in my career I was in the product
00:43:08
management and design side, so very hands on individual
00:43:11
contributor, but then was quickly leading teams.
00:43:14
So like I led the team, you know kind of as the general manager
00:43:18
on as a Camelot. I'm pretty early in my career
00:43:22
and then since then, so that would have been, you know, maybe
00:43:25
five years into my 12:12-ish year stint, I've been more on
00:43:29
the corporate executive side. So at, you know, war gaming, I
00:43:33
built and and ran the mobile division at App Love and I ran
00:43:38
the games business. And so as these things happen
00:43:41
like I became further away from kind of directly involved in the
00:43:45
creation operations and more on the business side.
00:43:48
And now you're now you're thrust back what?
00:43:50
Like exactly, Exactly. Two person, 1 1/2 person.
00:43:53
Yeah. Because I'll tell you, like,
00:43:55
there's nothing better in the industry it's than like actually
00:43:58
touching a game than actually, you know, touching consumers in
00:44:02
that way. You know, that's what I think we
00:44:04
all get in this industry for. And so I'm happy to be back.
00:44:07
And I I might poke at more what you're doing at the end, but
00:44:10
broadly? Fair to say it is like a game
00:44:13
development type approach rather than a particular game.
00:44:19
What do you mean by that? What you're what you're.
00:44:20
Working on now, are you building a game?
00:44:22
Like is your company a game, like building a particular game?
00:44:26
No. So we're we will build games but
00:44:29
our mission is more about ushering forth what we see as
00:44:34
this like new generation of games empowered by AI.
00:44:39
And so that's that's what our roughly our mission is.
00:44:42
It'll involve making games and. Evolve.
00:44:43
Yeah. We finally get into the picture
00:44:45
lines of technology, yeah. And and by the way, like I'm not
00:44:48
that us being stealth and and not sharing is not because I
00:44:51
don't want to. It's actually more as we I think
00:44:53
talked about or the call or before you started.
00:44:56
It's more that I struggle to concisely explain what it is
00:45:00
that we're doing. So you can so look, you open the
00:45:02
door. I'm gonna try to explain and
00:45:04
stumble way, way through. So yeah, nothing's off the
00:45:06
table. OK, cool.
00:45:07
We'll come back to that. Amy, can you give a sense of why
00:45:11
we think something's going to happen with artificial
00:45:13
intelligence and gaming or like what what you see, the promises?
00:45:17
Like, you know, in investing there can be a sense like, oh,
00:45:20
new things happened. It's an inflection point like
00:45:23
let's plug it into like our favorite thing.
00:45:25
But it does feel like with gaming in particular, there's
00:45:29
sort of a lot of optimism. I'm curious like what what you
00:45:32
think will make. What's is it what's happening in
00:45:35
generative AI with like? The image generation or what?
00:45:38
What in particular do you think is important here?
00:45:41
Yeah, I can break it down South. And also, you know, Keith can
00:45:44
give so many examples where he probably used AI in all of his
00:45:48
games. But AI in the gaming industry
00:45:52
has kind of been around since the beginning of games Like back
00:45:56
in, I don't know, fifties, 60s like starting with arcade,
00:45:58
right? Because you know, fundamentally
00:46:01
you're playing against. An opponent a lot of times and
00:46:04
before there was you're playing against a human you're playing
00:46:07
against a bot and and then with bots against other bots and so
00:46:13
that now is captured under kind of like you know non playable
00:46:16
character AI as definition. But this concept of vision gains
00:46:21
for multiple decades of time, and that's.
00:46:24
The NPC and probably when the most regular people or like you
00:46:28
know as a gamer you'd be like, the AI is like.
00:46:30
Sucks or like you, you know, famously I think like Mario Kart
00:46:33
where the AI just like is taught to like Slingshot to catch up if
00:46:36
you're too far ahead and all that.
00:46:38
Yeah. So we've all, we've all.
00:46:40
And like if you play a game too much, you sort of learn the mind
00:46:43
or whatever the rules are of the NPCS.
00:46:47
So yeah, that's an that's an exciting opportunity because
00:46:50
you're you interact with them and you sort of get to know how
00:46:52
they think. So that that's certainly one.
00:46:54
Sorry, God. Yeah, I know exactly and.
00:46:57
And I think, well we'll we'll hit the categories first and
00:47:00
then we'll go back into sort of the evolution of each of these.
00:47:03
The other, the other one is acid generation which also has had a
00:47:08
history in games cause back as far as you know there's like
00:47:12
Diablo or maybe even before the concept of procedural generation
00:47:16
of different levels. So you know, you're playing like
00:47:18
an action RPG and you're playing different like, you know,
00:47:21
dungeon levels and. And they're actually
00:47:23
procedurally generated. So you know, it might actually
00:47:26
look different for example, for different people each time you
00:47:29
go into a particular dungeon. Has been a concept that's been
00:47:32
baked into game engines for a very long time, which which I
00:47:36
think like the modern. I'd say like the post like last
00:47:39
fall like LLM version of these categories has been how do you
00:47:44
use large language models in these categories?
00:47:49
Because AI has been actually using games for a very long
00:47:51
time. And then the newest category
00:47:53
which Keith is, you know, building in is what what about
00:47:57
applying large language models to generating code that creates
00:48:01
games. So taking that in terms of asset
00:48:03
generation, one step further into actually game logic as
00:48:06
well. And people haven't done it yet.
00:48:09
And so it's like an exciting hypothetical space.
00:48:12
I mean, people are starting to build like basic games in that
00:48:14
space, but, and I think, you know, we'll get to it, but the
00:48:18
questions here are like one there is.
00:48:21
You know, a tremendous amount of enthusiasm.
00:48:24
I think some people in the games industry with a kind of like a
00:48:28
over hype of of the capabilities of AI, you know ahead of where
00:48:33
the actual technology is in some cases and in other cases people
00:48:37
say Oh yeah, this is the inevitable next step in
00:48:39
evolution of that. And that's one of the reasons
00:48:41
why you know AI devs are so excited.
00:48:45
Because there's no. The concept doesn't need to be
00:48:48
explained as existence industry for a long time, but I think
00:48:52
what everyone needs to remember is that you know, the bar for
00:48:54
creating content and games is so high and and new technology is
00:48:59
only like a part of that. It's a space where really
00:49:01
content is king. And it's just about like, you
00:49:03
know, can you play, can you produce better content?
00:49:05
That sort of fits into a big question for me, which is, you
00:49:09
know, I think there's this trend of like the democratization of
00:49:12
gaming or the sense that like you know.
00:49:16
Roblox or, you know whatever, you're trying to like maybe help
00:49:20
the user create games you're trying.
00:49:22
Or on the other hand, on mobile that there's just sort of a we
00:49:25
can make a bunch of games sort of cheaply and see what works
00:49:28
and then invest in that. Is, is that a true sort of
00:49:33
narrative or it At the end of the day, are most people just
00:49:36
flocking towards premium games that are really thought through
00:49:41
with like the best techniques, Like what What sort of the
00:49:44
optimism? For this sort of democratization
00:49:47
story in gaming and how much do you think AI is is going to move
00:49:53
sort of us closer one way or the other.
00:49:55
One thing that is easy to forget and and I am guilty of this as
00:49:58
well, even having been in the industry for all these years is
00:50:02
just how big and broad like the industry is, especially on the
00:50:05
consumer side, right. It's something like 434 billion
00:50:08
people. I don't know what the number is,
00:50:09
how play games. So that's like you know, sizable
00:50:11
portion of the human race. It is not like, you know, 20
00:50:15
years ago you would have said like when we said gamer, you
00:50:19
know, you thought of some like 30 or 20 something year old guy
00:50:21
in his mom's basement or something like that.
00:50:23
That's just not true anymore, right.
00:50:25
It's a very diverse market. Of and games are bigger than the
00:50:28
movie business and everything. It's like movies, music, TV, you
00:50:32
put it all together, games is bigger.
00:50:33
So it's like the largest form of entertainment that exists.
00:50:36
And so all of those kind of use cases that you mentioned are
00:50:39
real and are big. So hyper casual gaming which is
00:50:43
something that you kind of touched on as these like high
00:50:46
volume kind of mobile launches like that's actually a a large
00:50:51
category unto itself, something like 40 to 60% of the installs
00:50:55
across the mobile platform which is the biggest platform in
00:50:57
gaming or hyper casual style game.
00:51:00
But that's not to say like you still have Diablo releases like
00:51:03
billion dollar franchises that launch.
00:51:05
So all of that exists and all of that is part of game.
00:51:07
Nice, nice. When a business is big enough
00:51:09
that you're like, oh, we can go after one or the other, they're
00:51:12
both promising. I mean I'm curious like which
00:51:15
whether you think, yeah, the high end gaming or sort of the
00:51:20
sort of cheaper sort of more democratized game creation is,
00:51:24
is going to be helped by the trends that we're seeing here.
00:51:28
So I think all, all of it'll be helped but not equally, right
00:51:33
and not in the same way. So the high end, you know what
00:51:37
we think of as AAA but actually there's probably a bunch of
00:51:39
titles that we wouldn't consider AAA but meet this what I, what I
00:51:43
call it like the high complexity kind of side of the spectrum.
00:51:48
So that can, you know AAA is typically like high complexity
00:51:51
on the fidelity side, but there are games that are high
00:51:53
complexity in the kind of puzzle formation or other things that
00:51:56
are behind the scene or the multiplayer side of things.
00:51:59
So those those are all in the same bucket.
00:52:02
I expect those types of games. I expect for us to get more of
00:52:06
those games with AI, more of them produced.
00:52:08
So increase in productivity, increase in creativity via the
00:52:13
use of AI and then also new kind of features and components like
00:52:19
we talked about like a ILLM driven NPCS and then AI also
00:52:25
like natural language as an interface for gaming that will
00:52:29
be new as well. So I think that that kind of
00:52:31
thing is going to happen on that you know high complexity side.
00:52:36
I also think that things are going to change a lot on the
00:52:39
kind of democratization on the UGC user generated content side
00:52:43
of things as well. So some of it'll be similar like
00:52:46
natural language as an input, right?
00:52:48
Further abstracting and and thereby democratizing like
00:52:51
interaction with, you know, platforms like Roblox or even,
00:52:54
you know, more complex professional tools like Unity
00:52:57
and Unreal that will allow a lot more people to engage with the
00:53:02
creation of games. You know, from an intent
00:53:06
standpoint. In other words, have a game in
00:53:07
your head easier to make it playable.
00:53:11
But then I think there's actually, and this is where my
00:53:13
company is building tours, we believe that there's a third
00:53:16
category that gets born out of all of this, which wasn't
00:53:20
possible before and comes from the convergence of not just AI,
00:53:25
but also a number of other things, which allows for, yeah,
00:53:30
so for for for games and entertainment that look entirely
00:53:34
different than what we've seen before.
00:53:37
Well, OK. Different in what way Or give us
00:53:39
a little bit more there. Yeah, so, so, so here's where
00:53:45
all I've had to struggle through.
00:53:45
So like, the way I like to to come at this is from the player
00:53:50
perspective. So, like, there's been all these
00:53:53
things in games. That for my entire life as a
00:53:56
gamer and all the customers that that you know, millions of
00:53:58
customers that I've interacted with their billions really in in
00:54:01
the games that I've made all of these expectations that we've
00:54:03
had these unmet expectations and they keep being unmet and it's
00:54:07
just so disappointing and sad and so things like the fact that
00:54:12
games are largely A finite experience.
00:54:14
So even World of Warcraft still around right over that early MMO
00:54:18
I talked about the still around yes, it's true.
00:54:20
But like if you ask anybody that plays World of Warcraft, like
00:54:23
the amount of content that's being added, some people even
00:54:27
say it's dead. But like even that which kind of
00:54:29
pushes the boundary of finite still feels finite in many very
00:54:32
important ways and similar to that would be like unrealized
00:54:36
depth. And so you guys touched on this
00:54:39
a bit with like procedural generation and I think it was
00:54:41
something you said, Eric, where like a player can kind of learn
00:54:45
the mind of the NPCS and that gets boring on like this feeling
00:54:50
of alive. You know, Another another more
00:54:52
contemporary example would be Starfield, right.
00:54:54
So everybody is super excited for Starfield.
00:54:56
Oh my God, another Bethesda game.
00:54:58
Amazing, amazing. And it is amazing game.
00:55:00
But what is like, you know, one of the the often repeated
00:55:03
critiques is like, yes, there's 1000 worlds that can go explore,
00:55:07
but they all feel relatively the same they.
00:55:09
Feel right. Except for those people who
00:55:11
wanted to bring up Starfield as sort of the red flag for AI
00:55:15
gaming. Just because it's like, oh, we
00:55:18
can do everything, you know, we it can be infinite or whatever.
00:55:21
You know the sense that. I feel like more and more games
00:55:24
are, it's like, oh, maybe the way to win an open world is to
00:55:27
be even more open in world, you know, and and then and then it.
00:55:32
You lose sort of the story, right?
00:55:34
You hand it over to sort of assist, you know, some sort of
00:55:37
system to generate the world. But then you can't sort of put
00:55:41
enough human storytelling in to to power that expanse, I guess.
00:55:47
And that's how it's been with kind of you know, historical and
00:55:51
contemporary techniques of procedural generation and others
00:55:54
things we called AI before but are different from you know the
00:55:58
large language models, image models, things that are are
00:56:00
being used today and like with this also it's all so to
00:56:04
continue on the things, these unmet expectations like the
00:56:08
experiences are also relatively static.
00:56:10
In other words like there is a necessary one-size-fits-all that
00:56:14
goes into when you're developing a game, right.
00:56:17
I I can't customize the game for every player.
00:56:19
I can create systems so you know, you can choose and you
00:56:21
can, you can, you know, design your character, your player
00:56:24
character in in Starfield to some extent, right?
00:56:27
But you're still limited in your narrative choices.
00:56:29
You're still limited in, you know, the in in many other
00:56:31
things. And then because of the the
00:56:35
natural kind of formation of the industry and how games are
00:56:38
developed as well before AI, you end up with these fragmented
00:56:42
experiences. So like social is fragmenting
00:56:45
and everybody, like one of my favorite examples of this is
00:56:47
like you know, try and play a game with a one of your friends
00:56:51
on PlayStation, another one on PC or something like that.
00:56:54
Like shit, it's getting better, right?
00:56:56
But but damn, it is not great. And even beyond that, like if
00:57:00
you think about the kind of social interactions you can have
00:57:02
inside of a game, those really don't exist across games not
00:57:06
even within these you know small ecosystems like the PlayStation
00:57:10
Network or others that do try to create a social graph across
00:57:13
game. And then finally like what what
00:57:15
kind of summarizes all of this up for me and and speaks to you
00:57:17
know the direction that that we're going is games are not
00:57:21
personal, right? And I think that that that's
00:57:24
that's what we view as kind of this new frontier is a is an era
00:57:28
of personalized gaming experience.
00:57:31
So building games for individual level persons, for segments or
00:57:38
so not even I I actually think that on this you know this third
00:57:42
category of gaming that I was describing this native AI gaming
00:57:46
is not is going to look very different not just from a
00:57:48
product and consumer perspective but in a how you build a
00:57:51
perspective as well. Because, you know, creating an
00:57:55
AI game team, a game director, a personal game director who is
00:57:59
learning your preferences as you play.
00:58:02
Much like you know, Spotify is learning your preferences as you
00:58:05
listen. Or like the.
00:58:08
Game is building as you're going through it exactly and that's
00:58:13
and so, so things still have to get built in order to make that
00:58:16
happen. You must create you know
00:58:18
capabilities within a simulation that an AI, game director and
00:58:22
game team can work within on. But like the what we would have
00:58:28
traditionally called creating the game will be very different.
00:58:31
And I actually think that this one of my dreams or my hopes is
00:58:35
that this actually brings gaming much closer to what our earliest
00:58:40
and most free experiences with kind of interactive
00:58:44
entertainment probably were. Which was like hanging out in
00:58:47
the sandbox with your friends, right?
00:58:48
Where like there was no game director when you're in the
00:58:51
sandbox, right? There was no, like designer.
00:58:53
It was like, no, I'm, I'm the spaceman, you're the alien.
00:58:56
Like who? Who am I?
00:58:57
Shoot you? Oh no, I have a force field like
00:58:59
that kind of merging or blending or blurring between creator and
00:59:03
consumer. I think comes with this native
00:59:06
AI gaming future as well. One of my favorite books is
00:59:09
Ender's Game and you know in that in that book in that you
00:59:15
know he Ender is playing with a computer that creates basically
00:59:19
like psychological tests for him that sort of evolved the player
00:59:23
where you know the teachers don't even know where the gaming
00:59:25
is. I imagine that's still far away,
00:59:28
but it is amazing to imagine this world where.
00:59:31
The game is sort of responding to the player and producing
00:59:36
stuff as it goes. I mean that, you know, I, I I
00:59:40
think a lot from the framework of like TikTok, right, where,
00:59:44
you know, that's a key sort of AI sort of sorting, you know,
00:59:50
showing people what they want, running experiments based on how
00:59:54
people engage with stuff to see what which should rise to the
00:59:57
top. Yeah, I guess my, the thing that
01:00:02
I'd been hung up on with like longer form content is it just
01:00:05
like harder to run those experiments.
01:00:07
So even if like, you know, we're able to create like videos from
01:00:12
like whole cloth will, like we really have movies that are like
01:00:16
AB tested in that way because it's just like too hard to run
01:00:20
the experiments. What I hadn't really imagined is
01:00:22
this world you're talking about where like, Oh well, you have
01:00:25
somebody in the experience and you can sort of run many
01:00:27
experiments on them while they're playing.
01:00:29
The movie comparison is an interesting one and I think that
01:00:32
games in, in this case we actually have an advantage
01:00:34
because there is a lot more signal data that we get from a
01:00:39
person playing a game. Then you are probably going to
01:00:41
be able to get, you know, the person watching the movie.
01:00:44
Amy, could you take the temperature on where you think
01:00:48
startups are right now? I imagine you've been sort of
01:00:51
meeting with them like I don't know, yeah, what's what are you,
01:00:56
where do you, where are you seeing a lot of energy going and
01:00:58
sort of how, how, how active is this or like what?
01:01:01
Yeah. What's your view on sort of the
01:01:04
AI game space at the moment in Startup World?
01:01:07
Yeah. So now that, so now you know as
01:01:10
a partner at Menlo I've been covering games consumer and
01:01:13
games. But in games obviously you know
01:01:15
AI is a big part of it and I would say that the that the
01:01:21
where startups are and also I would say the.
01:01:24
Game studios of willingness to adopt right these tools and and
01:01:30
techniques really depends on where which of these kind of
01:01:33
categories that we we talked about the three you you have
01:01:36
like kind of MPC, you've got actually I'm going to swap the
01:01:39
order and we start with like asset generation creation and
01:01:44
then MPC and then sort of like cogeneration we are kind of in
01:01:48
that order of like I would say like readiness of adoption
01:01:51
because. On the on the asset generation
01:01:54
side, so this is, this is like taking image models and
01:01:57
generating kind of 2D and 3D images.
01:01:59
This is kind of like a where there's a lot of I would say
01:02:03
heat from artists because immediately they're like wow,
01:02:06
our our works, you know, our our works of art are being kind of
01:02:10
used to train these models without our permission.
01:02:13
And also you know are our jobs threatened?
01:02:16
And for some people, they are actually one of the studios when
01:02:19
I was at Lifespeed, the founder told.
01:02:22
Did tell me that, you know, they're actually able to reduce
01:02:26
the size of their 2D artist team by about 50% because you know
01:02:32
using a number of the tools out there whether it's like Leonardo
01:02:35
or Scenario or a number of these tools, they're actually and
01:02:38
actually the the number one tool that people use are actually
01:02:41
just mid journey and or building on stability, stable diffusion
01:02:45
directly. They're able to generate, you
01:02:48
know, let's say like concept art.
01:02:51
Everyone's trying out using these tools or concept art, but
01:02:54
actually getting some of the 2D assets to a point where it's
01:02:57
like you know 70% there and then taking it from there.
01:03:00
I think where the where the hardest problem is this 3D model
01:03:05
generation and the model or asset generation.
01:03:09
So the models really aren't quite there yet and so you have
01:03:14
tools out there that are a combination of.
01:03:17
You know, they'll they'll produce a model that say
01:03:20
anywhere from like 20 to 5060% of the way there.
01:03:24
And then they might actually have be mechanical turking like
01:03:27
the the remainder of the process to get you to like 70%.
01:03:29
And then your 3D artist takes that mechanical Turk, as in
01:03:32
there are people trying to flesh it out.
01:03:34
Is that what you're saying? On the company side, yeah, Yeah.
01:03:38
And nobody wants to say the word people, yeah.
01:03:42
No, I mean, I mean right now you cannot replace like a 3D artist
01:03:46
in terms of generating because the models just aren't there,
01:03:49
but because that is like the hardest and also the most cost
01:03:52
intensive, I think that will really unlock this piece of it.
01:03:57
Big question from game studios would be, well how do I protect
01:04:01
my IP? Because you know if I.
01:04:05
Will my IP be used to train this model, you know, not OK with
01:04:09
that? Or like if I generate using this
01:04:11
model assets, can I actually use them in my game and have it be
01:04:15
IP protected? And so I think this is a big
01:04:17
kind of area debate right now, and which could be helpful for
01:04:24
indie type developers, right? If if they don't have anything
01:04:28
to lose, they're willing to risk copyright more.
01:04:31
You know, it gives you a competitive edge.
01:04:32
You can be a little bit looser. While, you know, big companies
01:04:36
are certainly very worried about copyright.
01:04:38
Both. Both.
01:04:40
You know, whether they're going to hurt their own material, like
01:04:43
you're saying, or if they create stuff with these techniques,
01:04:47
whether they actually have any copyright claim.
01:04:49
Right, because there's some early signs that you don't.
01:04:53
Yeah. Anyway, that's a great point.
01:04:54
Yeah, absolutely. I mean, indie studios.
01:04:57
It's a big part of the thesis with this AI wave right now is
01:05:01
that. Cost content is the number one
01:05:04
cost in game making and so you can actually bridge this gap a
01:05:08
bit from for indie studios that are early that are like adopting
01:05:12
tooling earlier. I think over time it converges
01:05:16
And so then I think like in the end state, these tools will
01:05:19
actually just make it easier for all creators of they're all
01:05:22
artists to be creating these assets, which means that the
01:05:26
the, the artists within these large gaming studios actually
01:05:30
just be more efficient. And so I think it's still that
01:05:35
may still happen, but there's probably like an early adopter
01:05:37
advantage right now. But I think that's kind of
01:05:39
predicated on how good these models are.
01:05:42
And I think when it comes to putting assets into production,
01:05:45
I'm still not hearing almost anybody putting in production.
01:05:48
People are mostly using tooling on the concept side because in
01:05:52
terms of animating these these assets, particularly if it's 3D,
01:05:57
it's still, you know, we're still pretty far from that.
01:06:01
Yeah. So that's kind of acid
01:06:02
generation which is for this along adoption and right now
01:06:06
we're looking at founders that are enterprise ready, you know
01:06:09
knows how to make enterprise workflows and create them and
01:06:12
also go to market and and basically make a created
01:06:15
experience and also security and privacy that really scheme
01:06:19
studios need particularly on the larger side.
01:06:22
Do we see a foundation model for gaming?
01:06:25
Like has it? Have there been companies that
01:06:27
are trying to like do the Open AI thing specifically for
01:06:31
gaming? Or are they mostly just trying
01:06:33
to build off of Dolly Mid Journey and other sort of
01:06:38
general general tools? The image models are very good,
01:06:43
you know, in terms of what's what's out there and also like
01:06:47
you know, kind of developments on the open source front.
01:06:50
I think where a game specific model becomes interesting is
01:06:55
around generating code. And and the question there is
01:06:59
like, well, what are they training the data from?
01:07:01
Because you know, game code is, is, is basically IP that's held
01:07:06
within gaming companies, mostly large gaming companies that are
01:07:09
very incentivized to having that be in kind of like the open
01:07:13
public property so that other studios can actually use that to
01:07:17
generate games. And so I think that is like a
01:07:20
big open question, but current models really aren't trained to
01:07:24
be able to generate. Well, the neither the
01:07:27
specificity of code nor kind of like the complexity in terms of
01:07:31
knowing like well, what makes a fun game versus just like an
01:07:34
average not fun game. And because it's really like at
01:07:38
the outlier is where you actually have, you know, where
01:07:42
the fun is created and that's produces these large outcomes.
01:07:46
Yeah, all right. The, the non player character,
01:07:48
the AI in the game. I mean, I think it's such a fun
01:07:51
thing to think about for the reasons I was talking about
01:07:53
earlier. You know, people very much
01:07:55
experienced it. I mean, and and just given the
01:07:58
charm, you know, the fact that people think they're dealing
01:08:01
with a real being when you know, you're interacting with a
01:08:04
character bot or or whatever. Or like, you know, famously the
01:08:08
Google employee at Lambda like feels like once people are
01:08:11
interacting with three NPCS that have some of this LLM technology
01:08:16
behind it, it's going to be mind blowing.
01:08:19
Yeah, I I'm curious like Keith, like from game design, like what
01:08:25
are are we going to get like a sort of, I don't know like
01:08:29
what's it called Grand Theft Auto?
01:08:32
Is Grand Theft Auto or Red Dead or something going to have just
01:08:35
like totally free thinking LLMS like running around or do you
01:08:39
think that would break these games?
01:08:41
Or like how likely is it that somebody could just plug in like
01:08:44
sort of much more independent minded MPCS into sort of how we
01:08:48
think about traditional open world games?
01:08:52
Yeah, so personally I don't think the NPC, so I think the
01:08:58
NPC thing will happen. However, I don't think it will
01:09:01
be as or at least in the narrow way that I imagine most people
01:09:05
think of it, which is like like you were saying, like a Grand
01:09:07
Theft Auto and like all of the people walking around or a lot.
01:09:11
I think that that that will come probably not if you if you know.
01:09:15
The kind of development timelines at Rockstar and how
01:09:18
long they've been working on the latest one I would be shocked if
01:09:22
they kind of shoehorn it in here at the end but we'll see it in
01:09:26
games for sure. However I think that if you
01:09:29
broad how I don't think it'll be like that big unto itself
01:09:34
because the question there's all of these follow on questions.
01:09:37
So once you have like an LLM able to interact with the
01:09:39
creativity that we've seen in applications like ChatGPT but
01:09:43
within a game world the. Just interacting at A at A at a
01:09:47
language conversational level. Like, I don't know.
01:09:51
Like what freedom does the LLL have exactly?
01:09:53
What agency does it have? So I think when you start
01:09:55
talking about maybe a a more broad like kind of description
01:10:01
would be like an independent agents acting within the game
01:10:05
that are able to actually have an impact on the simulation
01:10:08
itself, on the gameplay experience itself more than just
01:10:11
like what we've what we're used to with NPCS.
01:10:14
Where like they might have a set of kind of actions that they
01:10:18
take, or if it's a bot that you're playing against, you know
01:10:20
it plays like you beyond that to something that is more material,
01:10:23
maybe more towards, you know, think of the difference between
01:10:27
an NPC and an RPG and a Dungeon Master in an in AD and D game,
01:10:32
right? Well then, well then that's when
01:10:33
it starts to feel like The Matrix.
01:10:35
You know, you've got Agent Smith in the game, you're shooting
01:10:37
against it. And then if the game developer
01:10:39
is empowered this AI system to like, you know change the rules
01:10:43
or in some way, that would yeah, be pretty mind alter or an easy
01:10:47
way to break games, you know, in.
01:10:50
Exactly. And and what I'd like to do is
01:10:52
actually link this to what Amy was just saying and and for all
01:10:56
of the, you know, developer like game developers out there and to
01:10:59
get a little hint on some of the technical solutions that we're
01:11:02
exploring right now, I totally agree with what Amy was saying
01:11:05
about generating code. It's something that we've been
01:11:07
experiencing as we've been testing out different models to
01:11:11
to try and generate code. But there is like.
01:11:14
I think this falls under the definition of code, but it's
01:11:16
different in an important way, which is data.
01:11:18
So like often times a game simulation is a set of code
01:11:24
which describes like which describes all the interactions
01:11:26
or which is interpreting data. And so a lot of game design
01:11:31
actually happens at a data level and.
01:11:34
LLMS are already so your foundational model, pre trained
01:11:37
LLMS, llama, GPD 4, whatever are already very good at data.
01:11:43
And So what do I mean by that? I mean like you know take a
01:11:46
game, Call of Duty right? There is like a whole bunch of
01:11:50
Call of Duty is code, but then a whole bunch of Call of Duty is
01:11:53
data and the part that's data is like the the configuration, the
01:11:57
data that configures the weapon that you have.
01:11:59
So you have an M4. And it's got like an attack
01:12:01
power of 50 in a range of like 35 that those numbers are data
01:12:07
and you so then you could easily say like come up with different
01:12:10
guns or something, yes. Yeah, exactly.
01:12:13
So that is already with it very much within the capabilities of
01:12:18
the models that we see today is the ability to take natural
01:12:21
language, understand game design, understand is
01:12:23
anthropomorphizing, but like to take natural language and
01:12:26
effectively translate it into. Functional contextually relevant
01:12:30
data inside of an existing simulation right.
01:12:33
So you could have way more skins like oh it generate us a bunch
01:12:36
of skins generate and and there where you have the constraints
01:12:41
you know it's not going to destroy your game.
01:12:43
Maybe you can you can review it before versus like my sort of
01:12:48
Agent Smith like wild futurist sort of thing is like OK, let an
01:12:52
AI rampant in in the game and like mess with it.
01:12:57
Yeah, Matrix is actually a great.
01:13:00
Example because like, you know, kind of foundational, at least
01:13:02
in the original Matrix was this idea of, you know, Neo as the
01:13:06
One and the Agents could break the simulation, right?
01:13:11
They worked outside of the simulation.
01:13:13
What we're saying here explicitly is like, yeah, sure,
01:13:16
that's cool with movies and all that, but like.
01:13:17
An agent working within the simulation can actually be very
01:13:21
powerful depending on how expansive your simulation is and
01:13:23
so to like wrap it back to the NPC.
01:13:26
Beyond just having a an interesting conversation, if an
01:13:29
NBC agent inside of a game has the ability to change data, that
01:13:35
can actually feel like pretty significant in terms of.
01:13:40
You know, a world feeling alive and things like that.
01:13:42
As a matter of fact, if you looked at like Starfield, right?
01:13:45
And I don't know this to be true 100%, but I would guess it to be
01:13:47
So a lot of what's going into the design of those worlds that
01:13:51
don't feel so much alive or the ones that do.
01:13:54
The big difference is the world's that don't feel alive.
01:13:56
Where procedurally the data was procedurally generated and the
01:13:58
world's that do feel alive were hand designed by people to feel
01:14:03
so, but all within the same simulation and so really just a
01:14:06
difference in data. Yeah, on the MPC side, part of
01:14:11
the conversation is that you gotta remember MPCS are pretty
01:14:15
good today. You know this.
01:14:17
This is what it players have been interacting with for the
01:14:19
last, you know, few decades. And so there are certain genres
01:14:25
of games in which kind of a IMPCS which is moving from like
01:14:28
a decision tree model for an MPC to a more open like kind of LLM
01:14:33
model for MPC is actually discernible to justify the
01:14:37
increased costs right around. Having that kind of deployment
01:14:41
would be in specific genres I would say like you know open
01:14:45
world and RPG being some of the most obvious ones.
01:14:47
And we're we are seeing, you know, new studios being born
01:14:51
with that approach. They're like, we're going to
01:14:53
build a game where we're differentiated by our characters
01:14:56
being LLMS, or they're going to sell LLM techniques to other
01:15:00
other people. So it's like so there's there's
01:15:04
a lot in between there and I would say there's yeah there's
01:15:06
like studios that are saying all right we're going to have these.
01:15:11
We're going to have a IMPC based characters that interact with
01:15:16
you that that actually will like enrich like where the the
01:15:20
stories might actually evolve like in in gameplay that starts
01:15:25
getting really closely tied to well how does like you know game
01:15:29
code also. Evolve, you know, dynamically
01:15:33
to, to Keith's point, and I think a lot of this theoretical
01:15:37
and what a lot of these game desks need to figure out is at
01:15:41
what, at what point do you give all those degrees of freedom?
01:15:44
And where do you actually just codify in the systems, which is,
01:15:47
you know, how game making is done today?
01:15:51
Because yeah, 'cause there's a cost consideration and then
01:15:54
there's also the controlling the player experience.
01:15:59
Part of it as well, you know if you actually have a a locked in
01:16:03
decision tree, you know like you are creating player paths for
01:16:06
these different segments of players and you're really
01:16:09
optimizing your their experience and you have a lot of control
01:16:11
there and where is giving up some of that control actually a
01:16:16
good thing? So I would say like, you know,
01:16:18
large studios when I talk to them on the MPC side are like
01:16:21
we're going to do kind of like proof of concepts.
01:16:27
Maybe with our smaller games, maybe with the new game and see
01:16:30
whether it actually impacts and improves the player experience
01:16:35
and then translates to like longer player retention and
01:16:38
higher kind of lifetime values and how much players are
01:16:41
spending in the game, etcetera. Before we make the determination
01:16:45
that it is completely game changing and we're going to
01:16:47
change all of our MPC systems across the games.
01:16:51
Yeah, yeah. I think this gives an
01:16:52
opportunity for a startup that wants to go.
01:16:54
A lot more further in this and and especially with their core
01:16:57
IP then where the incumbents are taking it.
01:17:00
So it's an exciting definitely. I can see the risks for like a
01:17:03
big developer. One like you, like you know with
01:17:06
Microsoft and Sydney or whatever, you don't you don't
01:17:08
know what the LMS will say. If you're just like sending out
01:17:11
a game and you like free response from NPCS could be like
01:17:16
terrifying. And two, I imagine it's got to
01:17:19
be like much riskier for just like creating bugs in a game
01:17:22
where you don't know the whole decision tree and you introduce
01:17:25
a part of a system that's basically like trying to come up
01:17:28
with a bunch of wild answers that you haven't gained out
01:17:32
before. But I mean, as a consumer, I
01:17:34
desperately want it. Like, you know, I feel like the
01:17:37
sense that like this game could come off the rails or that you
01:17:40
would have like an AI that's like feels Wilder.
01:17:43
Moving to our third category, I'm glad we have such like a
01:17:45
good framework like assets NPCS and sort of the the actual games
01:17:51
being and code generated by with help from AI.
01:17:55
Do we have like the first like AI game or like what are what is
01:18:01
the sort of state of play right now in terms of games, sort of
01:18:05
from whole cloth or components of games that are like human
01:18:10
fairly out of the loop? Expect us to begin on the the
01:18:15
game generation side. At the lowest level of
01:18:17
complexity possible. I mean it just makes sense to
01:18:19
start there. The funny thing is is actually
01:18:22
it you don't have to get very high on the complexity spectrum
01:18:26
before you start to be capable of generating a large portion of
01:18:31
the games that are actually played.
01:18:32
So if you think of you know categories of gaming especially
01:18:35
on mobile categories such as social casino, puzzle, I mean
01:18:39
puzzles like a multi billion dollar category social casino is
01:18:42
as well with with billions of players.
01:18:46
If you think of like 2D right is actually a low complexity game
01:18:50
on because it is like pixel graphics and Sprite sheets and
01:18:54
like 2 dimensional and all of that.
01:18:56
So like RPGs, top down strategy games, like again, like there's
01:19:01
actually a lot of gameplay that exists under that level and so
01:19:06
we might see the game revolution in like 2D gaming.
01:19:08
Like obviously mobile is popular right now, but also if we're
01:19:11
saying OK 3D assets are really hard, we want to democratize,
01:19:16
We're closer you know with with sort of yeah lower production
01:19:20
type games like it could mean a bunch of really cool like well
01:19:23
and you look at, you look at. Games slash platforms like
01:19:26
Roblox and Minecraft, right? That again, we're low complexity
01:19:31
on the fidelity still or largely low complexity on the fidelity
01:19:33
side, but we're able to go very deep on kind of the game playing
01:19:37
systems and customization and and creation and all of that.
01:19:42
Yeah, I think that that's where. A lot of the kind of, well, I'll
01:19:46
call it visible innovation will happen because of the problems
01:19:51
that like Amy was talking about earlier, right?
01:19:52
Creating 3D models, rigging those models like from a
01:19:55
generative perspective on visual effects, all of that shit.
01:19:59
I think there's entire teams of people at AAA studios that are
01:20:02
working on. Like simulating better, more
01:20:05
realistic like hair or movement and and stuff like that, right?
01:20:08
That's going to continue AI. Well, that's the classic, but
01:20:10
that's like every Pixar movie is like we've invented a new way
01:20:13
for curly hair to move. And like that's the technical
01:20:16
actually and underpinning and that's something is is
01:20:18
ultimately meaningful when it's accumulated altogether for sure.
01:20:22
But it's it that that type of advanced movie less obviously,
01:20:25
yes, I think on that the the kind of code generation side
01:20:28
within this within this three-step framework, you know
01:20:31
it's going to start at the low complexity side.
01:20:33
And So what we've already what we're seeing is we are seeing
01:20:35
some early experiments kind of hit around Twitter.
01:20:38
So you had you know people are using and actually I've done
01:20:40
this as well. It's one of the things that got
01:20:41
me really excited and I recommend anybody out there that
01:20:44
you know wants to dabble in this space, go do it is you can sit
01:20:47
down with Chachi BT right now and through like not one shot,
01:20:51
but recursively build the game. Like I've done.
01:20:53
There's a guy on Twitter with Yeah, I've done that.
01:20:56
I love that. Yeah.
01:20:56
Yeah. Yeah, yeah, yeah.
01:20:57
I played B&D with Chachi BT too. Yeah, exactly, exactly.
01:21:01
So. So that's already happening,
01:21:02
right? And then I'm talking even
01:21:04
generating code I tried to play. Poker with it too, though.
01:21:06
It's sort of like in in visual or like just chapter, just like.
01:21:10
You know, in the text format I got with D&D just to Stew on it
01:21:15
for a second, I the memory problems are what drive me
01:21:18
crazy. It's just like that it forgets
01:21:20
it and that you sort of have to try and remind it.
01:21:23
But it's so fun to have ChatGPT like make itself a character and
01:21:27
everything. Yeah, what would be fairly
01:21:28
probably easy is for you to spit out a a really like via ChatGPT
01:21:32
to spit out a fairly simple Python simulation or, you know,
01:21:36
pick another code. But Python seems to work
01:21:38
reasonably well. That sets up like the framework
01:21:41
for AD and D and then to use like GPT 4 as the kind of
01:21:46
language model within that simulation.
01:21:48
And you could do that and you'll have like you know you'll have a
01:21:52
a a better working like DMD kind of game that you created and and
01:21:57
so that's that's there. There's a few you know if you
01:21:59
want to check out FRVR and I know the the founders over
01:22:03
there, you know they've been working on this thing called Fr
01:22:06
VR4. They have some demos out some
01:22:08
videos out. I think they're starting to let
01:22:09
some people I believe they're a game film and some people were
01:22:12
able to play with it where it's actually like so natural
01:22:14
language prompt based you put it in and and you start to generate
01:22:17
like games playable like experiences and there are
01:22:23
there's a number of startups that are also like like like us
01:22:25
working on on this stuff as well and I expect in like the coming
01:22:28
year my my guess is next year we'll start to see a lot of
01:22:31
these you know hit more of a a broader production and and see
01:22:35
like what what they can do. Takes like three to five years
01:22:38
plus develop a AAA game, right. I think some of the more, yeah,
01:22:42
casual games will be faster. So like you know maybe they're.
01:22:45
I'm actually at on Real Fest in New Orleans right now.
01:22:48
Like seeing like what maps could be created.
01:22:50
With these kind of tools on UEFN, like that'll be
01:22:53
interesting cuz that'll take, that'll be faster to market.
01:22:56
Uni. What was it?
01:22:57
Uefn the What's the unreal Like Fortnite creator?
01:23:01
OK, OK. For Fortnite, yeah.
01:23:04
Do you have a? Do you have a sense just like
01:23:07
armchair of which? Big sort of game developers will
01:23:12
be most forward thinking or where where you see the most
01:23:15
aggression in terms of embracing what's going on.
01:23:18
I have a straw man. Obviously you know a lot of this
01:23:20
is like secret a lot of places, but my sense is that like a lot
01:23:24
of new tech Asia developers were faster to it.
01:23:29
So you know Netease tends to. I think both of the companies
01:23:31
are developing their own like large language and image models
01:23:36
and have been experimenting with all the tool tooling and.
01:23:40
You know they were some of their they were some of Asia
01:23:43
developers were some of the for like for a calm to us like we're
01:23:47
the first to kind of put higher quality I would say like you
01:23:52
know games watching games out live and and so I think they've
01:23:57
been they've been they've been moving and they've been
01:23:59
experimenting kind of for years in the AI space as well.
01:24:02
Keith, I mean, I wanted to just like come back to your company
01:24:04
for a second, like. I don't know, in two years, what
01:24:08
do you think? What is the first like thing
01:24:10
that you provide or like what's? I know there's a lot of sort of
01:24:14
ambition there, but just sort of in terms of something concrete
01:24:17
like what do you think the first move for you, you all is?
01:24:22
Yeah, so we're the first move might actually be AB to B
01:24:26
solution. I think there are many like
01:24:28
areas inside gaming right now where on you know our technology
01:24:33
can provide a a solve a very real problem that that some some
01:24:38
studios and publishers have that are creating or publishing games
01:24:41
of that low complexity threshold.
01:24:43
The thing that really excites me though and and you said two
01:24:45
years so, so that gives me more space to work in because I
01:24:47
expect next year for us to be executing on B2B is this you
01:24:51
know like lower complexity games that are creating themselves as
01:24:57
you play. I do think that we are you know
01:24:59
within two years of of us being able to launch games like that
01:25:02
and honestly we're not going to be the only ones.
01:25:04
So I'd expect to see a a number of games like that on the market
01:25:07
within two years. And you, well, you'll be
01:25:09
designing some of it. You'll be like, oh, it's a
01:25:12
Subway surfer game, but the AI is free to do this thing.
01:25:16
Or you think from sort of the beginning in some way the AI is
01:25:20
the way that we're doing it is we are designing kind of the the
01:25:22
foundational capabilities. It's sort of like you know, if
01:25:25
you think of think of chat JPT right and GPT 4 and and language
01:25:31
models right. They are attempting to predict
01:25:33
the next word, right? And and what has been just so
01:25:37
like so unexpectedly, at least for me like surprising is that
01:25:41
in trying to predict the next word when you feel like throw
01:25:45
enough parameters and like training data and all that stuff
01:25:47
at it, you end up with what feels a lot like intelligence
01:25:50
and like something that's emergent kind of capability.
01:25:53
And so for us it's about predicting the next gameplay
01:25:56
experience, the next moment of gameplay, the next level and I
01:26:00
would expect similarly so. So like you know we're testing
01:26:04
our systems with trying to figure out hey can it make a
01:26:07
match three-game. Can it make a a level of
01:26:09
something that looks like Mario Brothers.
01:26:11
Yes and yes. So you have to build the.
01:26:13
First level and then you're like, what's the next?
01:26:18
Yes, actually that's part of, but not for.
01:26:21
We have to build a level. So yeah, you're really fast,
01:26:26
like technical under the hood. So like we have what's called a
01:26:29
game descriptor, which is a data structure that describes the
01:26:32
game in data or describes a game in data.
01:26:35
And then is is run through an interpreter that actually
01:26:37
produces the game. And then we use right now Unity
01:26:40
to to actually render and interact with the the
01:26:42
application. And so in that like framework or
01:26:45
schema, like we produced the first game descriptor, the first
01:26:51
data that describes a game within the simulation.
01:26:53
But then and and let's say that the first thing that we built
01:26:56
was like a something that looked like a 2D platformer like a
01:26:59
Mario or build Prince of Persia's.
01:27:01
But then we said here's a game descriptor of Mario to the LLM
01:27:05
now make us a game descriptor for a top down shooting game.
01:27:10
And it did and it works. And so like that's like it's
01:27:14
it's like in a way if that if that answers your question
01:27:17
right, like yes in a way you need to populate it so that it
01:27:20
understands the simulation and the parameters that it has to
01:27:22
work with. But it's not like I have to make
01:27:25
the first level of all the games.
01:27:27
I just make a game and it functions and works and then all
01:27:30
the other games can be generated.
01:27:32
Amy, I just wanted to ask like a lesson or two from like the
01:27:36
crypto gaming experience or like moving, you know, I I think on a
01:27:41
super surface level take it's like.
01:27:43
There was a lot of excitement in crypto and now there's a lot of
01:27:46
excitement in AI like will you know, some of it overhyped or
01:27:49
like I don't know what. What are the lessons from having
01:27:52
gone through sort of the heat of crypto and now sort of and
01:27:57
thinking a lot about sort of AI stuff now?
01:27:59
I'll, I'll I'll kind of keep my answer to the crypto blotching
01:28:03
game side just because I mean the broader crypto space.
01:28:06
I would say that the the token launches and the abuse of that.
01:28:11
You know, really kind of dictated a lot of that
01:28:12
narrative. But on the gaming side, I to be
01:28:16
honest, I I always thought there was like way too much venture
01:28:19
dollars within the blockchain gaming space.
01:28:22
There were a lot of really bright developers that embarked
01:28:27
and wanted to make games for the first time, which is really
01:28:30
difficult. I mean, games are really
01:28:31
difficult to make, right? Not only do you have to know how
01:28:34
to make a complicated system work that balances the game and
01:28:39
produces. You know, kind of our thousand
01:28:42
hours of content for people to have fun with, but also how to
01:28:45
live operate that afterwards. I mean this is, it's just really
01:28:48
difficult. And so it took some time before
01:28:52
I would say experienced game devs really actually came into
01:28:56
the blockchain industry and was excited as well in terms of
01:28:58
finding ways to leverage blockchain technology that that,
01:29:03
you know, could make gameplay feel like something that a
01:29:06
player hasn't experienced before.
01:29:08
I mean if it basically feels the same, then.
01:29:10
I would say anything that adds friction actually it's just bad
01:29:13
in game design but it has to be it has to be part of like it has
01:29:18
to create a gaming meta that that is actually totally
01:29:23
different in in that and is unique And so I would basically
01:29:28
apply and so if you look at all the VC dollars put like that
01:29:32
went into the blockchain games industry a lot of it was not to
01:29:35
fun teams with a clear vision of it was more just like.
01:29:39
I'm going to fund it because it's in this category, but the
01:29:41
devil is absolutely the detail, especially with the rate of
01:29:44
failure in the game's content side.
01:29:47
And so I would apply the same kind of, you know, view on with
01:29:53
with with the AI intersection which is 1.
01:29:56
Like you know, just having the content be AI generated does not
01:30:01
mean anyone is going to find this font, and so the bar is it
01:30:05
needs to be really great content.
01:30:07
And also it needs to be and compared to what the other
01:30:13
things that I could play in the market, because you know, the
01:30:15
most valuable thing is actually person's time.
01:30:17
You know, why would I play some game when I can just play Breath
01:30:21
of the Wild or Diablo Or, you know, CS?
01:30:24
Investors may want to invest because it has sort of a new
01:30:27
technique, but at the end of the day the players don't care that
01:30:31
much how it was created and want it to end up being a great game
01:30:35
and. The people who can do that, or
01:30:36
often the people who've built great games.
01:30:39
Before players don't care about ideology, they care about and I
01:30:43
but I'm game to ask you sometimes, but I mean players
01:30:45
just care about like one of the content is just fun.
01:30:48
All right. Our last question what looking
01:30:50
forward five years like with the promise of today, like what what
01:30:55
is something that you're most excited about what might be
01:30:58
possible in in gaming and AI. When I look back five years ago
01:31:02
in gaming what I'm most and then I look forward five years.
01:31:06
What I'm most excited about is for me personally and I think
01:31:10
many people within the industry this last five years have been
01:31:12
kind of boring just straight up like as a as an industry
01:31:15
participant you know it took five years to make Starfield and
01:31:20
again I'm gonna play it, I'm gonna enjoy it.
01:31:23
But you know I I doubt that I'm gonna be blown away by it.
01:31:26
And that's really been the story just over and over and over.
01:31:29
And the truth is if I ask myself and I look at the, you know, the
01:31:32
the best games that were launched in the last five years,
01:31:35
none of them were, you know, none of them were unlocked or
01:31:39
made possible by some specific technology that came out in the
01:31:42
last five years. There's a lot of same same,
01:31:44
there's a lot of sequels and like that's honestly
01:31:47
unfortunate. And it's not just limited to
01:31:48
games. It's been all over, I think
01:31:50
entertainment and media, when I look forward to five years, what
01:31:52
excites me is that I think that picture will be different.
01:31:56
And I also think that a lot of the terminology we use describe
01:32:01
gaming is going to get so mushed and blurred and blended that
01:32:07
it's just going to look very, very different.
01:32:08
I think what what is a consumer or a player, what is a creator,
01:32:13
what is a a streamer or a passive participant.
01:32:17
I think all of those lines are going to get blurred in really
01:32:21
exciting ways as AI enables you know a a a change in how we
01:32:26
interact with technology in a democratization to kind of the
01:32:29
interaction with technology and and that unknown is what excites
01:32:33
me. That's unknown is what I'm I'm
01:32:35
striving for and and whether I'm right with the company that
01:32:38
we're building or not like I'm I'm here for the change I I.
01:32:42
Something new, Yeah, I I totally agree with that sentiment.
01:32:45
And yeah, the blurring that feels so inevitable.
01:32:49
That's a exciting vision, Amy. What would what would you offer?
01:32:55
Yeah, I would agree. So I think in five years most of
01:32:59
the game comments will still be the same game comments.
01:33:02
Will there be one or two new entrants that has launched
01:33:06
something pretty incredible leveraging you know the new
01:33:10
tech. I I certainly hope so.
01:33:13
And I think where I hear, where we hear like skepticism,
01:33:16
incumbents still will both make us like, you know, kind of
01:33:22
proceed cautiously because you know.
01:33:25
And that's so much knowledge is back there.
01:33:27
But also, all right, this is not where a lot of people are
01:33:30
willing to risk, you know, most $9 billion titles.
01:33:32
So this gives an opportunity for a startup, an indie studio to
01:33:37
really break out. So yeah, I I'm hoping there will
01:33:40
be a couple of of sort of AI native studios that are able to
01:33:43
create something super awesome. Awesome.
01:33:47
Keith, Amy, thanks so much for coming on the show.
01:33:49
This is great. Thanks so much.
01:33:51
Thanks, Eric. All right, fun talking about
01:33:53
games. Thanks so much.
01:33:55
That's our episode. I'm Eric Newcomer, Co host, our
01:33:58
Max child and James Wilsterman Co founders of Volley and my
01:34:02
long time friends. Thank you to Scott Brody, our
01:34:05
producer, Riley Kinsella, my chief of staff, Gabby Caliendo,
01:34:09
key person of Volley who's helping make this whole
01:34:11
conference happen. Shout out to young Chomsky for
01:34:15
the theme music. Please like, comment, subscribe
01:34:18
on YouTube. Follow us and give us a review
01:34:21
on Apple Podcasts. And of course, subscribe to
01:34:24
newcomernewcomer.co, Become a paid subscriber, support
01:34:30
everything we do, and then go play Volley games on your Alexa
01:34:34
say, hey Alexa, let's play Song quiz anyway, All right.
01:34:38
Thanks so much. See you next week.