Transcript
Use with ChatGPT / Claude Auto-scroll A
The hosts introduce the episode’s focus on AI as a major technological shift in software and games. They explain why the topic matters now and why they want to focus on timeless implications rather than fast-changing tools.
Speaker A
Foreign.
Hello, and welcome to the Game Craft Podcast.I'm your host, Mitch Lasky.
Speaker B
I'm Blake Robbins.
Speaker A
And this is a podcast about the modern history of the video game business.Episode five, Thoughts on navigating the AI future.
Speaker B
Oh, man, this one, this one.I feel like we gave the teaser in episode one, but, man, I'm excited to do a full episode on this.
Speaker A
Hopefully this one will be a little more rigorous than the intro to episode one, but look, we've been pretty critical of the games industry backlash against the use of AI in game production.And in this episode, I think we're going to make an attempt to expand on our argument that AI is a technological innovation with significant implications for the games business.And it may be bigger still.
Right.It may be a real paradigm shift in the way that software and media are constructed in the modern world.And even before going on, I want to talk a little bit about what I mean by paradigm shift.I think a lot of people sort of pay lip service to that.Etymologically, it goes back really to the early 60s when Thomas Kuhn, the historian of science, wrote the Structure of Scientific Revolutions.
And in that book he talks about changes in the way that a whole mindset around a problem space changed.And so, for example, when the Copernican view of the universe replaced the Galilean view of the universe, and all of a sudden the sun no longer revolved around the Earth, but the earth revolved around the sun, or the discovery of oxygen, where before they thought that there was this substance in flammable objects called phlogiston that burned, and then suddenly they realized, no, there was a chemical reaction going on, and these things completely changed the perceptual way of approaching a problem space.And I think that AI in software is really worthy of being called a paradigm shift in a certain way because it does require complete change in the way we approach the problem space.So we wanted to discuss the parts of the game's business that we think will continue to be valuable and investable in an AI era.
And we also wanted to identify some of the AI tooling that we are aware of, that we've seen in some cases that some of the companies we work with are using, beyond the obvious ones like Codex and Claude code, that I think everyone is aware of.
Speaker B
Yeah, I think it's also worth.I mean, just your paradigm shift is 100% real.I mean, you and I feel this in our day to day of just where we sit and we're watching how fast the traditional software world that you And I interact with completely change.But I think it's worth, like, two disclaimers.One is, like, this world is changing faster than ever.
Who knows by the time we're done recording this episode, what's been released or announced?Like, we're talking, the rate of change in this world is unlike anything I've ever seen.And then the second is, there's also everything that we're talking about is the worst it's ever going to be.And so whatever capabilities we're talking about of AI in production or in game creation or any of that, it's the worst it's ever going to be.And the past few months, we've already seen that to be true of just how fast this all continues to expand.So I just want to say those two real quickly, before we jump too far.
Speaker A
You had to provide a sort of tangible example of that.In February, I wrote a primer for my family on, like, how to use code, how to use Claude code, and how to use Claude cowork.And I put a little disclaimer in it saying, like, probably don't read this after, like, March 30th.And actually that turned out to be perfectly right.Like, after March 30th, everything changed again.And, like, now some of some different tools are sort of better serving that particular use case.And again, that was a period of maybe 45 days.So this stuff ages like milk in the sun.And I think you have to just take that under consideration and really get yourself up to speed.If you're listening to this episode in the future as opposed to in real time, everything may very well have changed.
Speaker B
But even despite all that, we actually created this episode or drafted this episode with the idea of what's actually going to be timeless.And so that's part of why I'm so excited to do this one, is we tried to think much more of what are the things that aren't going to change or what are the responses.Responses to the critiques.And so I'm excited to really dive in.
They lay out the episode’s intent: responding to dismissive takes, warning about competitive pressure, and pushing back on AI maximalism. The discussion emphasizes that AI is already changing software production and will reshape the games business.
Speaker A
Cool.Well, let's talk about the intent of this episode.Since intent is an important part of getting AI to do stuff correctly, we'll take that same path in terms of talking about introducing this episode.So part of the intent of this episode is really an effort to respond to what I really hear as this frequent critique from artists and programmers that I know who've done what I call AI drive bys, where they decided for a couple of days or an afternoon or maybe a week to try and get some work Done.They fumbled around with the tools, they didn't really get what they needed out of it, or they spent the time and it didn't produce the results that they wanted.
And from that experience, that drive by experience, they've sort of come to the conclusion that AI is dumb and that it's just another crypto metaverse level distraction that the industry doesn't need to worry about or even should resist.And I think we both really fundamentally disagree with that position.And we really think that this is real.And I would say the.Look, here's what's real.
GitHub disclosed recently that commits are up 1,400% year over year, 14x year over year.Now, that's current run rate, right?And it's actually accelerating.And their expectation is that it may actually be higher than 14x year over year for the full year.Now, GitHub is not a startup.
GitHub is a repository platform that is operating already at insane scale.I mean, they were already.I mean, it's something like 275 billion commits a day or a week.I can't remember what the number was, but I mean, we're talking about one of the most scaled platforms in existence, and it increased in volume by 14 times.Now, you know, a critic might say, well, that's just a bunch of amateur programmers throwing AI slop at the wall.But you know what?That's not possible.There's not enough amateur programmers out there to create that level of growth.And so this has to be.The vast majority of this increase has to be professional and enterprise usage.
So look, that is real.And there's no way you're going to get to that kind of growth without the code being generated primarily by artificial intelligences.
Speaker B
Yeah, I mean, look, it's just even anecdotal on that.You have the anthropic team saying that almost all of the code for Claude code, which is their main product that's sort of out there on the enterprise side, is being all written with AI.They're literally like the main guy, Boris is on podcast saying, I can't remember the last time I hand wrote any of the code.I'm talking with my agent and then it's writing all of this for me.And so.And you see this time and time again, if you talk to any founder in Silicon Valley, they're like, yeah, we use AI on 70% of our code, 80% of our code, and most of the ones that aren't even at 100% is just because the way they architect it and they're working to get to 100% of their code.So give it six months or a year and that might be 100% true.
And so, yeah, I mean, even, and we were talking about this yesterday, it's like, even if you take the most skeptical view and you're like, oh, it's only 10% of this is real, which you and I both think that's insane to think.
But still, even if you do that, it's like, okay, it's still up 140% year over year, which would be bonkers at the scale that GitHub is at.
Speaker A
Yeah, it's crazy.So look, the fact that the majority of game developers, and we'll get to the majority of game developers later, because there's some interesting data back from GDC from some surveys that were done of game industry professionals at the show, but the majority of game developers are essentially shunning what the rest of the software industry is adopting, like in earnest.And that should make us all a little bit worried, right?So that's number one.
Another part of the intent of this episode is to sound an alarm bell about competitive necessity.You know, in our Istanbul episode, the last episode that we recorded, and in our prior season episodes on China and on the globalization of the games business, we've hopefully expressed that the historical Western and Japanese dominance of the games business is really being challenged in new markets and from new markets.And most of those markets don't have moral issues with AI.If, and I think, as we suspect is true, AI is going to be a long term transformative technology, refusal to learn and embrace it is really going to magnify these competitive threats that we're already facing from new markets, from new entrants into this market.So I think that's another intent of ours in recording this.And then I think it's also a response to the AI maximalists.
So don't get it wrong from us.We're not here talking about sort of let's just surrender to our AI overlords.I think there's a group of people out there who don't know much about the games business, but who know a lot about AI, who think that the games business is just going to turn into basically write a prompt and it'll, you can get a whole game from a prompt or you know, nonsense like that.And I think we don't believe that.
And I think frankly, if the current state of things is any indication, making these technologies work in a professional development environment to create real entertainment experiences, whether linear or interactive, is no small feat.Like, it takes a lot of orchestration, it takes a lot of work to get the tools to do what you want them to do.Now, when you get it right, you can produce magic.But it is a struggle.And I can tell you, just having gone through this in my Hollywood production company, just trying to get scenes that are coherent out of these things without visual hallucinations and other things is a struggle, but it can be done.And when you get it, and when you get those workflows working properly, the productivity increase is just insane.And the ability to make really interesting content without the same cost structure is absolutely there and valid.
Speaker B
Yeah, and look, this might sound bad, but so much of our industry, and let's say especially mobile games, is built on almost arbitrages.And it's hard to think of an area where you can achieve these arbitrages by just simply being a little bit ahead of the curve in this current moment.And so it's just another piece here which is, I don't know if you're a Turkish studio and I've talked to many, they are for sure embracing all of these things and they're for sure tinkering in how to do UA and make new ads and all those pieces.And so there's flaws, but it also will be rewarded if you take the time to go and actually push through.
Speaker A
Yeah, and it's frustrating again, as an amateur on the sidelines trying to use some of these tools and understand them.Look, the LLM frontier models are moving targets, right?I mean, we went from opus 4.6, from anthropic and we went to 4.7.And in some cases, because they didn't have enough compute to handle it, it kind of regressed a little bit and it stopped being really as good at some of the two things that it was good at before.
In the meantime, OpenAI released GPT 5.5 and you know, on the high reasoning setting and the fast setting, it's really, really performant where 5.3 wasn't really that good.So, you know, again, these things are moving targets.You gotta be incredibly nimble, really have to kind of get in and understand them as a native in order to make them work.But again, when you do and you get the care and feeding of them properly aligned, they can really do wonderful things.So I think finally, and I think most importantly, what we want to talk about is not just sort of what's happening, but also what we think is going to happen in the future.
What incumbent companies, platforms and skills are going to survive or even thrive in this new paradigm.The power of the Frontier Model LLMs and the workflow harnesses to disrupt existing work is clear.Right, but where are the spaces that these models can't reach that won't be that easy for the LLMs to swallow?What is going to have its value and importance magnified in this new paradigm shift that was perhaps undervalued before?And I think this is going to be both good and bad for certain parts of the industry as we will discuss.
I think people who think well, like maintaining the status quo or whatever is the secret to resisting the intrusion of AI into the game space.Be careful what you wish for because in many ways it may be, and I think we'll argue that it will be in many cases that the incumbents and that the power centers in the games business that we are not entirely comfortable with currently are going to get even stronger and more powerful in an AI enabled future.
The hosts define the central thesis: AI will hollow out the middle of game development while making top-tier developers, publishers, and platforms more valuable. They argue that the deeper question is which layers above the model will control value creation.
Speaker B
Yeah, yeah, look, when we were talking about this episode, I just thought your framing of what's going to endure was just a really smart way to approach this overall topic.I think you and I could sit here and obviously talk about AI tooling and talk about how you and I think it's going to change a lot of things, but I think it's more interesting to be like what's not going to change or what's going to get stronger.And so, yeah, I mean maybe we just talk about what is the overall thesis.
We've sort of danced around it a little bit, but what is the thesis?
Speaker A
So I'll propose, and you can hammer on it I think, but I think the thesis that I'd like to operate under for this episode is that AI is going to hollow out the middle, the big middle of the game development cost stack and it's going to leave the top tier of developers, publishers, creatives and platforms more valuable than before.And this may have the effect of further eroding the opportunity for middle tier games because it's going to essentially pull the ladder up to a certain extent when it comes to distribution, funding, publishing opportunities, et cetera.And so that's the broad, I think that's the broad philosophical thesis and I think breaking it down a little bit, I'd say we're going to see a vast, diminishing, vast lessening of the costs of attempts.
Speaker B
Right.
Speaker A
So I think, you know, you and I have talked about this quite a bit and I think you have a perspective on this, but like you're going to get a lot more shots on Goal for the same cost.The ability to iterate before you commit to full production.
Speaker B
Yep.So it's going to be, I mean the speed.Not even, not even.Just like price or resource allocation, the speed to have attempts should collapse closer to zero than like I don't know you and I know current timelines are, you know, not even, we're talking not even two, three years ago, companies were raising, you know, $30 million to go and release a vertical slice in two years.That is if that is still your mindset at all.You were just wrong.Like we are not going back to that world.The speed and the number of attempts you can have in this current moment is bonkers.
Speaker A
So while the cost of prototyping is collapsing, the cost of actually engineering a hit may not be.And it may end up being just as expensive to take a hit to market, to fully produce and market and distribute a hit as it has been in the past.But the hope would be that you'd have a lot more certainty about audience acceptance and the ability for that hit to actually be a hit rather than the kind of throw wet spaghetti at the wall that we currently experience in a lot of go to market in the video game business, as evidenced by things like Concord.
Speaker B
Do you think that's true?Where my own interpretation of what you just said is you and I, my default assume costs will go down on the production side.
But whatever those costs might be, let's say overall budget might stay the same because it might get filled up in marketing or those other areas.Is that how you're interpreting the cost of the hits made or you think even production stay in the same budget?
Speaker A
I'm looking at it more as a percentage of overall spend.Right.In the sense that let's say you had 30 million available.I mean, I'm thinking that in the AI enabled world and I think we're already starting to see this, so this is really not that much of a projection because I already hear venture capitalists talking about this to their portfolio companies, which is that in an AI enabled world, you're not getting the 100 million, you're getting 30, you know what I mean?Through a couple of rounds.And so essentially you're going to need to make that.You're going to need to make what you would have made before for 100, for 30 and also get it to market.And so while we may see efficiencies actually enabled by AI, and we'll talk about some of those in terms of the customer acquisition and marketing spend, you're still going to end up not really saving money, if you will.
Maybe, yeah, you'll save the money that you didn't get, but you're not going to save the money that you did get.
Speaker B
Fair.Yes.Yes.Cool.
That's perfect.
Speaker A
And so, so that's number one.Number two, I think there's.We're going to talk about this in much, much greater detail, but we've identified basically five what we'd call durable layers of the games business that we'll explore later in this podcast.The places that AI won't be able to reach as easily.
And I think one of the things we're going to talk about is the fact that those areas are particularly favorable to companies that already have users, IP and data at scale.
Speaker B
Right.
Speaker A
So maybe less favorable to startups and more favorable to incumbents.And then the final part of the thesis is that the surface debate that AI will take creative jobs, flood distribution channels with swap, replace junior engineers, is actually real.I mean, we're not Pollyannish about this.We don't think that this is some kind of miracle that's going to lead to full employment.And whatever it's like, this is going to disruptive as all paradigm shifts are, as paradigm shifts have historically been, even in the video game business itself.But those concerns are distracting from the deeper question, which is who's going to control the layers above the model?And it's those layers above the model that I think we're going to spend the majority of this episode talking about.Because those are the durable layers and those are the things where value is really going to be created beyond just cost savings and efficiency.
A thought experiment shows how cheap game creation could become for individuals and small teams. They argue that while prototyping costs collapse, market forces and distribution constraints still shape what becomes valuable.
Speaker B
Yeah, I mean, you just covered what I perceive to be like the real game craft take on AI in games.I think a lot of people obviously talk to us in private about how are we interpreting the current moment.And so I feel very good about like this as a rough outline to go through and talk about our current stance on a lot of this.
Speaker A
Cool.So let's start with a thought experiment, right?And it's just, it's a little goofy, but we'll do it anyway.Here's the scenario.You got a 17 year old girl on a single weekend in April of 2026 sitting in her bedroom.Vibe codes a working, you know, first person shooter prototype in Uefn, maybe Roblox generates the core 3D assets using Meshi or cube dubs the Voice, work with 11 labs and writes gameplay scripts in cursor or Claude code and ships a playable build before Sunday dinner.
Speaker B
Right.
Speaker A
That is not a Fantasy.Right.That is a possibility.In the current world, you'd need skills, but you wouldn't need extraordinary skills.So given that that's a possibility that didn't exist until really this moment, the question isn't really any longer, will AI affect games?
Right.I think that's an absolutely ridiculous and specious question at the mom.Of course AI will affect games.The question's really, if all that development is now cheap, what is valuable?
Speaker B
Yes, yes, that is a real scenario.
I don't even know if it's a thought experiment in that we could probably go and join a random discord server where kids are building in Roblox and AI or UEFN and AI or AI and games.And you could probably see some amazing, amazing things that people are making and they would all probably say they're doing it after hours after their job.They probably work some random job.And like you said, development is now cheap.We can debate what that really means, but at the end of the day, we should assume that the cost of one of these games rounds closer to zero than certainly the $100 million budgets they were getting before.
But I think it does really open up.Back to that question of, okay then, where does value actually accrue?If we have infinite games that are literally any idea that you want to bring to market, if that's always, and this is maybe a controversial aside, I actually don't know if there's been a shortage of good games to begin with.You know, this is sort of the like, overall topic that you and I have discussed, which is like, there are already probably too many games in the.On Steam, there are already too many games on the app.
Speaker A
I think that's the ironic thing here because you hear all of this criticism about AI leading to, you know, oversupply and slop.I would argue that we already have oversupply and slop without AI.
Speaker B
Yes.
As determined by the voting machine or the weighing machine of the market.Yes.
Of where dollars are allocated by consumers.Some long tail person might say, no, no, we need more of this specific style of game.Well, the market is saying they really care about these top five games.And so it doesn't really matter if you flood it with another 100,000 games or a million games.There might be some stuff that trickle through, but the core of how games break through or accrue value, I think largely stays the same in our framing.
Speaker A
Yeah, but I think it's worth interrogating what changes when costs really collapse in this way.Right.Because I think traditionally, and we've been arguing this since episode one of the Gamecraft podcast four years ago, which is that market forces drive and constrain a lot of the creative choices that people make in games.And they don't.They pretend that that's not happening, but it's happening.Right.It's like you make games in established genres because you know you can sell them.
Speaker B
Right.
Speaker A
You make games with particular kinds of art styles because you think that that's what your audience wants.You make games of a certain length and duration, games with monetization strategies, games with certain pricing target audiences and make them for certain platforms.
All of those choices are market driven.Whether you like it or not, they're all market driven.Maybe in some sense you're deciding to be contrarian and you're making a choice that's not consensus.And of course we know the sort of Bill Gurley theory that non consensus correct is the box you actually want to be in because that's where you produce usually the highest returns.But cost has always been a market driving constraint in the video games business.And when that starts to disappear, the ability to make different choices, the cost of those different choices may really change considerably and that equation may change considerably.And I think that's the exciting part of this.
Speaker B
Yep.By the way, one very simple example of this is we should almost assume any game that gets any meaningful traction can be cross platform on day zero.And that like the ability to port games nowadays is like, okay, like yeah, what?
There you go.It's going to be on every platform, port and localize.
Speaker A
I mean, you know, because that's the other thing.
Speaker B
Right.
Speaker A
It's like, okay, so before we get into what survives part of this, I do think we want to at least talk a little bit about the fact that the backlash against AI that we're seeing in the video game business and that, you know, we heard from people at gdc, et cetera, isn't totally illogical.
Speaker B
Right.
Speaker A
We mentioned this before.
Speaker B
Right.
Speaker A
This is going to be really disruptive.We're not here to sugarcoat the risks to the current cohort of artists and programmers and designers because those risks are real and they're probably inevitable that they're going to produce dislocation and disruption.
But that's what happens when technology innovation creates a paradigm shift.I mean, we saw this when 3D games and ubiquitous GPUs kind of came into the, into the picture.It didn't wipe out the market for 2D artists, but it changed their utility and value in the ecosystem in a meaningful way.It increased the value of 3D programmers and 3D art skills modelers who didn't really exist in the 2D world of sprites and textures and things.So mobile similarly didn't wipe out core gaming genres, but it did open up the industry to competition in particular markets where core game patterns and platforms were not as competitively solid.So for, let's say older people or casual gamers who wouldn't go through the pain and suffering of buying a CD ROM or downloading something from Steam or even creating a Steam account in the first place, but they'd play an iPhone game, for example, or emerging markets like Turkey, like the Middle east, like other developing markets around the world where access to consoles and PCs was a cost prohibitive idea and therefore the mobile phone which you already had became the ubiquitous gaming device.
So yeah, it didn't.It was additive rather than necessarily like subtractive in that way.But all of these paradigm shifts are disruptive and they are.And they all affect the value of labor because they change the value proposition for skills.
Speaker B
Yes.Yeah.I mean it's just driving home exactly what you said before, which is just games are driven by the markets and the market driven constraints is what drives everything.And we're experiencing this massive paradigm shift as you said.And so I really want to underscore just what you said of like 3D didn't wipe out 2D artists, it just changed their utility and value in the ecosystem.
By the way, weirdly, and this is maybe nostalgia or whatever almost come full circle where maybe in this current moment 2D artists and 2D art style is maybe the highest it's ever been since its previous peak.And so I don't want to also they all have their value in their place.But it is just important to say how these changes impact things in the short and long run.They all have their overall role, but of course the utility and values going to change.
They address backlash from developers, labor concerns, copyright worries, and accusations of AI slop. The hosts argue these objections are understandable but ultimately miss the larger market and platform dynamics.
Speaker A
But the rub is that.
Let's get to the GDC 2026 survey because I think this was super instructive in this regard.36% of game surveyed attendees admitted to using generative AI at work, which in my opinion is dreadfully low.Right, 36%.
Speaker B
I mean you obviously have sat on many boards, you have ran your own companies.If you were running, I don't know, let's just say JAMDAT in current moment and 36% of your engineers or your team was using AI right now.
What would your response to that?
Speaker A
Oh my God, I would be burning the boats.Right?I mean that's the way I ran the company to begin with, it's like we were the most successful.This is a forgotten technology.
But there was this wireless application protocol called wap, which was basically like the HTML of mobile phone, pre iPhone mobile phones.And we had like a multi million user WAP game called Gladiator that was like incredibly successful and that we could probably have turned into some kind of limited advertising play or whatever.And upon the launch of J2ME and Brew, which were coding platforms that let you make applications, we saw there was a future where we would be able to actually make standalone games that were not going to be bound by this hideous WAP technology.And we literally just incinerated all of the success that we had had to date and said, we're moving 100% into this new world.And I would have done the same thing today if I was running a studio, right.I would basically be like, okay, I mean, I may not go to the same degree that some of these corporations we're seeing are going, where they're basically rewarding people for burning tokens because I was always a bit of a cost hwk.
But I would say that I would be embracing this with 100%.So I think beyond that, 36% that said that admitted that they were using generative AI in their work.52% said that it was having a negative effect on the industry.
Speaker B
A majority of people said it had
Speaker A
a negative effect on their team.
Speaker B
Negative effect.Oh, man.
Speaker A
And it's.
Speaker B
Yeah, I mean, it's hard to reconcile that with just the other world that you and I are in of just like the normal software world.And that like, the person who is maybe at most risk of having their job displaced isn't even the artist, it's actually the software engineer in the current moment.Right.And yet, like Silicon Valley, the founders and teams that we're talking to, the people that are most excited about it are the software engineers.And so it's a little bit of a paradox.And they're like, what?These are the people who should be the most concerned that they're going away.And it's like, no, no.They all feel the efficiency gains and they all see like, oh, I don't need to go and recreate this actual system setup and built a compiler, all of it.And so that's the part that's a little harder for me.
Speaker A
But I think it's worth trying.
I think we need to try and understand it.And I think the reason that I.It's important to think about why this is happening is I think it's important to understand why this is happening because the industry has been through a series of crazy dislocations pre AI, right?Like moving into this, like with AI kind of coming at the end of this period of, you know, going back really many, many years to like, you know, the labor disputes that happened between labor and management over, let's say, overtime and the categorization of software workers as managerial and therefore not subject to overtime.And then they could.So that they could be worked to death during crunch time.And the settlement with Electronic Arts and all the things that happened way back where there was a lot of exploitation of labor in the games business.You know, you had all kinds of dislocations moving into this current environment with the funding of games drying up and the pre AI slop of trying to get a game noticed, an indie game noticed.
It's like there have been.It's just beating after beating that sort of.
The lower rungs of employees in the video game industry have taken over the last decade and a half, two decades, frankly.And given that, I think AI coming in and sort of delivering the coup de grace is understandable.Right?It's understandable why someone would have this response to this and why 52% of respondents at the, at GDC would have said that it was having a negative impact on the industry.Again, I think that is a perfectly rational, if wrong take on AI.
Speaker B
Well said.
Speaker A
So look, I think there's been.Let's just quickly list the objections and go through them one by one.We won't spend a whole lot of time on this.But like labor displacement, 28% of GDC respondents in that survey said that they had been laid off in the prior two years.28%.Now, we've heard numbers as high as a third, and I would argue that perhaps some of that third that I think is probably the more appropriate number.They didn't even show up at GDC to look for a job.So they weren't even included in the survey.
Speaker B
Right.
Speaker A
The 28% who said that they'd been laid off in the last two years were still looking for work in the industry.So I think that number could even be higher than.There was a little bit of kind of survivorship bias, if you will, in that 28% number.So look, developers connect AI with things like outsourcing, with things like riffs.And I think that's not necessarily been the case broadly in the video game business to date.But if you look out in the broader software industry, there's companies in Software that have been very upfront about the fact that they're doing significant layoffs because of AI efficiency.And so it's not again, irrational for programmers and others inside of the video game business to view AI as threatening in that regard.And I think that's fair.We've seen SAG and AFTRA on strike for 11 months in 2025 over this interactive media agreement primarily where there was a lot of AI language, right.
A lot of stuff in there about the restrict restrictions on the use of likenesses of digital replicas, how those were going to be compensated, reporting on usage of AI in actor replacement or voice replacement, et cetera.
And those were hard won rights.And the strike ended.And I would argue in a lot of ways, and I've had this experience before, it's like I was going to
Speaker B
say, weren't you involved with.Didn't you advise on.
Was it Writers Guild?
Speaker A
Yeah.So in 2007 the Writers Guilds were going on strike and I was asked by a couple of the strike captains to just sort of quietly consult with them about their technology strategy, sort of the future of technology.And I got on the phone with them and I said, here's what I would do if I were you.And I outlined some fairly in your face negotiating tactics regarding future technologies.
And they came back to me and they were like, like, no man, we just need, we just want an eighth of a percent more on DVD sales.This was in 2007, right?And I was like, look, with all due respect, it's like, I appreciate that but like, you need to understand that like DVDs may not be long for this world and this was on the eve of streaming and they were so backward looking and worried about now in the new agreement, I would say they've gotten a little bit more religion.And some of the, the people that were involved in the pre strike negotiations in the ultimate collective bargaining agreement that came out of it, like John August for example, are very technologically sophisticated.But those people weren't around in 2007 when I was working with the WGA.And I have to say that they consistently play defense.
They don't really play.It's like there were probably ways that they could have negotiated a much more interesting and forward looking sort of use of their digital Persona.But instead it was all about, no, no, no, you can't do this, you can't do that, you can't do this, you can't do that.Instead of like, how is this a new market for my likeness and how can I profit from the use of my likeness.
Speaker B
Yes, yes.And look, unfortunately these markets are efficient and consumers get what they want on some level or the markets demands it.And I mean you're seeing it with the latest Sea Dance video model and things of like people will just abuse it or take these likeness even if it's against these type of things.And look, it's just a game of whack a mole on some level.And I don't support that necessarily, but it is a natural response to the market.And so, yeah, I sort of understand the defense, but also the world has changed.We are in a paradigm shift.And so you need to think more of like, like what's the next 10, 20 years gonna look like?Rather than how we solve the current moment.
Speaker A
Yeah, I think that's exactly right.
Another objection that we hear consistently, and I hear this mostly from lawyers, which I'm always a little bit suspect of because I went through this again earlier in the video game business in the 90s when the guilds, the Hollywood guilds were sort of saber rattling about the fact that they wanted to get involved in the video game business.And all the lawyers that I knew were scrambling around trying to be like, oh, we gotta go sign deal, we gotta get EA to go sign a deal with sag.We gotta get EA to go sign a deal with, with the Director's Guild or whatever.And it's like that was, it ended up being complete nonsense.So I'm a little bit, whenever I see the lawyers coming in and saying, oh, this is the thing we need to worry about, I'm always a little suspicious because it's like these are things that are going to make more work for lawyers as opposed to more sort of more general value for the industry.
But anyway, you know, there is a larger question here in the way that LLMs were trained and how that will have copyright implications potentially going forward.Right.Obviously these models were not trained exclusively on open source data.They were trained on proprietary data as well.And so the concept of copyright and the concept of authorship is still very, very much to be negotiated in terms of some of these things.Whether or not there's going to be any resolution in my lifetime, I'm not entirely sure.Maybe in your lifetime, I don't know.
Speaker B
I mean, look, you have such a interesting view because maybe listeners don't know the full context of.You were a lawyer, intellectual property lawyer.And so I imagine you're sitting at a seat of like, wow, this is the stuff we used to fight over is.
I don't even know it's so tiny relative to this.This is like people stealing full catalogs of movies and full on uploading them and training on them and Yeah, I mean, I'm not nearly smart enough on the IP law side of like what is protected and whether this is actually protected.
Speaker A
Yeah, whatever.Without getting into the weeds of this stuff, it's like I would say that, look, we had the sort of, the similar thing happen in the early days of social media and of the video on the Internet and all these other things.And we ended up with the dmca and for all its warts and all its flaws, the DMCA basically got everybody to chill and the Internet took off as a media experience.
Speaker B
Right.
Speaker A
And I think it's very likely that we're going to see something very similar happen here.
Speaker B
Right.
Speaker A
That we're going to figure out some kind of a negotiated settlement that's going to probably not get everybody exactly what they want, but is going to at least provide a layer of certainty where we know we can utilize these models within certain parameters.And that's exactly what happened with media on the web.Right.The DMCA basically came in and it addressed piracy and IP rights and other things like that in linear media.And YouTube and TikTok just took off within that legal framework.It's like, okay, good, you know what?You can't upload full movies that you ripped off of, you know, whatever pirate server you are, you are working on and post them on YouTube.You can't use copyrighted music in your videos now.Yeah, look, a lot of people complain about that.And if you, and if you've ever gotten a takedown notice for inadvertently having a song on, on the radio in the background While on your YouTube video, you'll recognize that, that this is, you know, it's pretty draconian.
But you know what, if you can operate within the legal framework, you can do fine.And that's, I'm sure going to happen here too.So I don't think this is, I think that this is a bunch of sound and fury signifying nothing from a group of lawyers who stand to profit from a lot of litigation around this.And at the end of the day we're going to have either a legislative or legal court driven settlement to this problem and we're just going to go on with it and we're going to understand what the boundaries are for sure.We talked about AI Slop and how AI Slop is just meet the new slop, same as the old slop.I think the market solves this problem.And I think the market's currently solving this problem.
Speaker B
Yeah, it will reward the good content and the slop will just either be a thing that people meme and make jokes about or it just goes away completely.It's not going to get attention.
Speaker A
Yeah.And look, we are going to talk about this a little bit later in the next section when we talk about one of the things that we think survives.But I think curation is going to be of even greater importance.
And I think it's something that some of the more open platforms like steam and the iOS App Store, which are currently in a slop situation with 20 plus thousand games annually coming to market, could do a much better job of this and may actually be rewarded for limiting access to content from and curtailing some of that activity as opposed to continuing to foster it.And then finally, this whole concept, this last bit of which, this really offends me, which is this sort of disclosure as a scarlet letter.This is the Steam approach currently, which is let's shame the people that are using AI in, in making their content.And I think what's so absurd about it, I mean, look, it's as misguided and hypocritical as it was in Hawthorne's novel of the same name, right?That 90% of Google Cloud reported that their game industry survey indicated that 90% of games had at least some AI code in them.
Right.And it's invisible if you didn't disclose it, it.
Speaker B
Right.
Speaker A
There's no way anybody would figure it out.It just looks like code to, you know, there's no way you could really tell, at least not from playing the game.
Speaker B
Right.
Speaker A
But where you can tell is if there was, if there was, you know, generative AI art or generative AI sound or music or voices or whatever.And so it's that 10%, right?That's, that's, that's a, that's sort of the lightning rod for all this attention.Whereas like everything that's going on under the surface, this 90% of just code generated by AI is largely invisible.
And I think this disclosure thing is just stupid.And it's really, I think long term, not in the interest of Steam or these other platforms to basically pander to this Luddite community and make this an issue of morality around whether or not you're using AI or not using AI in the generation of.As if somehow you're a better or worse person for having created.It's like we should be evaluating these things on the quality of the games.
Speaker B
Is this pure virtue signaling it would be like the App Store having a disclaimer of when I clicked on the Uber app of like, they used AI to make your app great.I don't care.Does the app work?Is it a great service?Okay, that doesn't mean anything to me.Of like, oh, wow, DoorDash uses AI.Like, wow, I don't care.Why do I care about the individual game if they use that for coding?
It's a lot of trying to appeal to the crowd maybe is the way to frame that, which is they're just like.A lot of people are angry and upset for a whole bunch of reasons.Maybe some of what we just discussed, but.But to just make it so you're singling out what is inevitably going to be 100% of games using AI seems really short sighted.Again, we should just assume by default that all of them are going to use it in the same way that an accountant used Excel once they moved to that, yes, that's going to be a thing.
And so we're just going to assume every game is going to have disclaimers on it now.
Speaker A
All right, let's get to the five categories that survive.
The first durable layer is trust, including identity, certification, payment rails, and platform verification. They explain why trusted platforms and user-aggregation systems become more valuable in an AI-heavy world.
Speaker A
So let's just jump in, I'll tell you what they are and then we'll go through them one by one and you can tell me where I'm wrong about some of these and we'll get into it.
Speaker B
I'm excited.
Speaker A
Here are the five.
The five things that survive are trust, context, distribution, curated taste, and liability.Right.Those are the five categories, I think, that not only survive the AI paradigm shift, but actually are enabled and thrive post AI paradigm shift that are even more valuable in the future than they are in the present, where they're already pretty valuable.And the TLDR here is AI is going to amplify some of the trends that we've been talking about on this podcast for the last four seasons that we think are the really significant aspects of the industry.I think those things are going to become even more significant.
Speaker B
Yeah.I mean, look, and we'll get into each of these, but a lot of this comes back to the platforms, which is what you and I have been talking about from day zero, which is actually try and build a real business, actually try and build a platform, actually aggregate users, not just try and build a single game.And I don't know if a single game studio has almost any of these traits.And so I'm excited to go into detail.
Speaker A
Cool.
I think I wanted to give a quick hat tip to Nate Jones, who is where I kind of got the original idea to think about this because he put out a YouTube video in which he talked about this in a more abstract and fashion, talking about software broadly.It wasn't exactly.It had nothing to do with games, but I thought his framework of how to think about it in terms of what actually is defensible and what's not defensible, what LLMs can reach and what LLMs can't reach was really interesting.And I kind of mapped his idea, if you will, onto the games business.And that's sort of where these five categories come from.
So let's start with trust.And what do we mean by trust so broadly?This is like certified apps and services, payment trust, age verification, federated identity.AI cannot manufacture credentialed trust.And so any repository of credentialed trust, again, whether that's economic, whether that's personal identity, et cetera, is going to be outside the reach of the LLM and therefore the owners of that and the people who can profit from its use become more valuable loci in the equation.So mapping that onto games, the app stores, broadly, roblox particularly now that they are implementing age verification.Same with Discord, the ESRB Ratings and Anti Cheat, you know, payment rails.We've talked about all the sort of human touching aspects of this where it's like, prove to me that you're human.Or by proving that you're human, you have certain entitlements that you don't have as an agent.
Speaker B
Right.
Speaker A
As an AI software agent.And so I think all of those things become increasingly valuable.
Speaker B
Do you include like IP and that side here?Or does that fall more in the curated taste in context?
Speaker A
I think the, I think AI and sort of the IP end of it is going to, I think really fall more into context than into trust.Although obviously it straddles it a bit.Because you trust the brands, right?
Speaker B
Yeah, yeah, yeah.That's why.Yeah.
In the same way when you go to a grocery store, you might buy the name brand over the other.
Speaker A
Exactly.That is an aspect of trust.But I'm talking more here about sort of like proving that you're human, human.Right.
And supplying attributes of your humanness to systems in order to get certain entitlements.
Speaker B
For sure.Another example here would be like stripe on the payment side of.Okay, this is a trusted way to go through this, maybe Visa, MasterCard, whatever you want to use as the example of that is going to continue to endure and be valuable in this next phase.
Speaker A
Absolutely.
So who are the winners in that case?In the Games business, it's the user aggregation platforms.Because the user aggregation platforms have all of this, you know, they have this data and they can either federate it or they can use it internally.So you know, you have Steam, the app stores, the console certification.Like the ability to certify, again, an aspect of trust is the ability to certify that games are that they have a source of origin that is authentic.
So for example, that a game that's labeled as a game from Electronic Arts or that's called FIFA or whatever, back to your brand trust question is actually coming from that particular source.Right.So I think that's an important aspect of this.And also the ability to certify that software is virus free, you know, is relatively bug free.That the consoles.It's a role the consoles have historically played.
Speaker B
Right.
Speaker A
Or Apple has played where you have to submit your game for certification and it goes through a testing process so that the end user has a certain kind of good housekeeping seal of approval.Right.That's an aspect of trust as well.And I think those platforms that are able to sort of produce that trust and aggregate it are going to become really increasingly valuable in this new world.
They define context as the data, systems of record, and story universes that give AI meaningful grounding. In games, this includes player behavior, community data, IP constraints, and personalization memory.
Speaker B
Yeah, it's funny, my mind almost also goes to like they're going to play such a role in what gets approved as the gatekeepers.Not even in the curation standpoint, but just from a, like, there is trust in that this is a game, this is not malicious.In a world where you're flooding the app stores or these aggregated platforms, there's a lot of trust that's going to be baked in.And the content disclosure stuff is one example of this and maybe to the negative, but it is zero attempt of trying to instill trust onto the platform.
Speaker A
Yeah, I think it is a signal that they see themselves as a trust broker.It's a misguided, in my opinion, use of that trust brokerage, but I do think it is directionally correct.
Speaker B
Yeah, they were trying to wield their stance in the market of how do we continue to put our finger on the scale?Maybe because players were asking for that could be.
Speaker A
Okay.Number two is context.And so what do we mean by context?We mean.And again, context has a very specific meaning within the AI world in terms of sort of like loading context.Right.It's like establishing for the AI the parameters of the problem space that you're going to then interrogate.So we mean it.
We're sort of using that definition a bit and also sort of broadening it a little bit.And I Think so the things that would fit in this bucket are like proprietary systems of record right.Outside the games business.For example corporate data lakes, notion databases.Right.
Like other kinds of personal and corporate database repositories that store systems of record, social graphs, behavioral data, expert content, for example, like you know, content that is unique to a certain community, the skills that are unique to that community.And I'd also include story universes in there, right?Because they provide for example Lord of the Rings, right, Like thousands of pages of context, right.That could be used for a variety of purposes.And so all of those things are really interesting in the games business.
We're going to see things like LiveOps, telemetry, player behavior monetization, data community, CRM.Again, store universe is I think really important.IP catalogs and creator and modding communities.Right.I think all of those are going to be sources of context that are going to become extremely valuable if they can be aggregated and if they could be utilized and deployed in this future.
Speaker B
Yes.I mean I think of context in the simplest way as like which games or platforms have the easiest ability to reconstruct who they think you are or who do they think their player is?And the thing that's going to be increasingly valuable, that sits on top of the actual models themselves is memory.And oh, Mitch really likes going on win streaks or he churns the moment he goes on three game loss streaks and almost player personalization on some level.And so all of these things are going to get supercharged in an AI world.
And the context of like who is this player?What type of player base do we even have?Who are they friends with all of those things and a lot of this ties together, but even just their age and all of those pieces really come together and to supercharge the AI and that's what they're really good at, is help customize experiences.
Speaker A
There's a whole bunch of different potential use cases for this.But I also wouldn't ignore the fact that things like Cannon Rails for example, for IP are an important kind of context that's going to be really valuable because again, if you have a UGC platform with AI enabledment that is capable of generating new stories in existing universes, you're going to want the context of what's allowable and what's not allowable.
We've certainly read enough fan fiction and we've read fan fiction that tries to color within the lines and we've read fan fiction that take certain liberties with the, you know, let's say sexual orientation of the characters or whatever.And I think those kinds of things, being able to own and proscribe them is going to be an important aspect of context in an AI world.
Speaker B
Yep.
The discussion turns to distribution, including storefronts, social platforms, search, and UA channels. They argue that discovery and traffic control will matter even more as content supply explodes.
Speaker A
So let's move on to distribution.Our favorite topic.
Speaker B
The King.The king still.
Speaker A
So look, the general way of thinking about this broadly in software is storefronts, social platforms, marketplaces, code repositories.All of these loci of distribution are going to be, I think, elevated in a way that we can't even currently imagine.Because ultimately, at the end of the day, finding stuff is really, really interesting.
I've had interesting conversations lately with some really smart people about what the future of SEO is.Search engine optimization.When you're searching effectively through an LLM in chatting with an LLM and that LLM is frozen in time and then has to go out to the web for current context, how do you influence that search in the way that you have historically done with SEO?And it's a really interesting question, and it's a question that's still as yet unanswered, but that's an important aspect of distribution.SEO has been vital to getting traffic to websites and being able to show up in high page rank on the Google algorithm, for example.
And we have no idea what the equivalent of that page rank is going to be.I have certain theories about what that might be and I'm encouraging companies that I'm working with to think about SEO very differently than they've historically thought about it.But again, who would have thought that we would be in a position now where something that we've taken for granted for the last 25 years is going to be greatly devalued.
Speaker B
Yeah, I mean, look, a simple example is like go to ChatGPT and ask what game you should play right now and just see what's there.Like, you know, and like that game, whatever is at the top of the charts or top five there, they are absolutely getting traffic.And you know, what game should my like 12 year old son play?Or you know what, like all of that is going to be what has historically been dominated on the Google side of filling up blog posts and whatever else.There is a whole new world that people are going to navigate.I think the other side of this is, in my mind, it just makes the applovins of the world that much more valuable.Absolutely.
Speaker A
I was going to say the UA enablers are going to be super valuable again because if there's this ability to get people, particularly if we do see what I'm hoping we see, which is a Lot more boutique e commerce experiences in games.Right.And we'll get to it in the next section when we talk about curation.But I think one of the antidotes to this slop problem is boutiques, right?It's like, okay, yeah, you don't like going into a department store and seeing 7,000 T shirts.
Instead you can go to somebody who selected like 10 beautiful colors and styles for you, right.And I think, and you'd maybe even be willing to pay a premium as you often are in retail to buy from a boutique because you're paying essentially partly for that curation.And I think we're going to see that really have an effect on distribution because in an atomized world of distribution, which may actually happen, where there's many, many more potential avenues to get games to market, being able to control that surface like Applovin does, of sort of getting people from the ad to the product is incredibly valuable.
Speaker B
Yeah, I sort of hinted to it at the beginning of this episode, but I almost think it's like you can imagine if a game raised $20 million and it might only take 5 million to do it.They might have to spend the remaining budget just on UA and marketing to break through the noise in this current moment.
They explore how taste, orchestration, and editorial judgment become more important when anyone can generate content. Publishers, boutiques, and highly curated platforms may gain influence as the middle market gets squeezed.
Speaker A
So let's turn to the fourth category and that's curated taste.And I would include in that orchestration.Right, because it's a little bit abstract in the games business, but in the broader software world orchestration is becoming really important as a, a means of enabling your products to talk to agents directly as opposed to having human intermediation.And there's a thesis that orchestration quality is going to be the equivalent of curated taste for agents so that they're going to look for the least friction in the ability to implement those things into their orchestrations as opposed to it's sort of what the human equivalent of curated taste would be.Which is kind of a fun and interesting thought experiment.But this is the editorial function, right direction, judgment, creativity, workflow, expertise, hard won expertise that puts the right tools on the right job at the right time.
And I think this is in the games context, really, really important because this is really the part, part that I think that people in games shouldn't be afraid of.Right.Because I think this creative design hero, asset creation, feel, voice, art direction, editorial taste, even at the platform level, all of these things are going to be more important in an AI enabled world because they're the things that AI can't do or can't do well or they do in a way that's very obviously not human.And you know, people will say they're getting better at it.It's.The fakes are getting better, the deep fakes, the, you know, your ability, the voice mimicry.I've certainly seen people, you know, unfortunately there's a lot of context out there of my voice on the Internet and people have done some fun parodies and you know what, they're, they're quite good.
But they're not me.
Speaker B
Right.
Speaker A
And they're obviously not me.And I think that that human component is going to be persistent.
Speaker B
Yes, I think there's on some level this is historically been more of the IGN reviews or Metacritic or whatever site you would use, but I think there's going to be a lot more of this editorial driven, especially on the indie side.My Twitter feed is still filled with this is a really cool new indie game that's popping up or I'm starting this thing.And I think it all comes down to taste, whatever that means.
Taste might evolve in this new world, but look, the games creatives will always be truly creatives and so I think largely pretty immune to this.Yeah, I agree as a whole, especially especially the indies, I think more than anyone.
Speaker A
I mean this is probably an unpopular take amongst the AI maximalists, but I would say that the entertainment broadly like not, you know, packaged kind of, you know, TMZ type entertainment or you know, formulaic soap operas or maybe even games in spaces like Shooters or Match three or four Forex that are almost programmatic at this point.But a lot of other forms of entertainment kind of have a natural immunity to AI, right?Because they.
Speaker B
I completely agree.
Speaker A
The kind of pleasure that people are seeking in them is very, very different and is maybe not accessible at least or may not be accessible immediately and I think maybe quite defensible.
Speaker B
I couldn't agree more.
Speaker A
I think the converse however is also true, which is if you're making mediocre, non innovative paint by numbers content within extraordinarily well known and well trodden genres, you may be in trouble.
Speaker B
Yeah, well, I mean look, and we obviously just did a whole episode on turkey and casual puzzles and I think that is a world that really is around efficiency gains and hyper optimization.
And look, for better or worse, AI is amazing at that and AI is going to level that entire genre and industry up and competition is going to be crazy.But I don't know if that's a place where taste in the traditional sense that you and I are talking about will shine through.And I think that's Actually a place where it's much more calculus of fun in all aspects.
Speaker A
Yeah.I mean, look, when we articulated the thesis of this episode, we talked about the fact that this was going to produce a hollowing out of the mid market.
Very likely.And the mid market is filled with this kind of content.
Speaker B
Yes.
Speaker A
And so if anything, that is going to accelerate this wipeout of the mid market.
Speaker B
Right, yeah.
Look, the dreams of the world on the Royal Match and Match three side will actually be fine, but it's everyone else, the Candy crushes will be fine.
Speaker A
Yeah.Because they have context, because they've aggregated users.Hopefully.Hopefully.
Right, yes.But the value is not necessarily going to be in that hyper optimization any longer because that's going to be increasingly thin margins.
Speaker B
Yes, yes.Yeah.And maybe said differently is like the floor of what a good game is or the expectations of a game might also rise and so the middle might get even more blurry because it's just like everyone has these personalizations, everyone has these optimizations and so to stand out is that much harder.
Speaker A
Yeah, I think it's akin to what we saw early on with the coding harnesses, where I think this articulation was really correct, which is that the LLMs basically are a little better than an average programmer.
Speaker B
Right.
Speaker A
But the idea is that that means that they're better than 50% of programmers who are already out there.And so in a sense it is a kind of like a sort of floor raising technology in that way.
Speaker B
Right.
Speaker A
It sort of raises everybody up to average.And if everybody gets raised up to average, like average, average is no longer Good enough.
Speaker B
Yes, 100%.
Speaker A
So I think, look, this sort of curation and taste thing I think also has a very interesting sort of subcategory, which is that when everyone can generate, when everyone can create, the scarce resource is knowing what is worth making making and maybe what is worth promoting.
Speaker B
Right.
Speaker A
In the case of some of these stores.And I think one of the things that I'm really looking forward to seeing is whether or not the Apple App Store in particular and Steam become more highly curated because they've taken the position that they're really not going to be super curated.Right.It's like you go into a Barnes and Noble in the past when they existed and there would be a shelf that would be like, like staff picks.
Speaker B
Right.
Speaker A
And there'd be like somebody'd write a little blurb underneath it.You know, it's like, why this particular Toni Morrison novel or William Gibson novel or whatever was their pick for science fiction and it was wonderful.It was like really cool.You got a little insight into the person and you got a little insight into the book or whatever.And it's like that doesn't exist really in games outside of social media.
And I think it's incumbent upon the platforms to start reintroducing things like that.That.Right, like just a feature where they could have celebrity or famous players or whatever, have their lists.
Speaker B
Right?Yes, yes.
I mean, look, you obviously remember the times of like when an app store feature from iOS or Apple mattered a big deal.
Speaker A
Exactly.
Speaker B
Today it doesn't matter.Maybe it was a couple of weeks ago you and I tried to go find the top 100 games and like that was buried.That used to be the true curation.
That took maybe 30 seconds, 45 seconds of clicking around the app store to find it.And so weirdly, they are maybe extremely behind or have taken an extremely hands off approach on curation.But I am with you in a time where it's just going to get flooded more and more.It's never been more important.
Speaker A
I agree.
And again, I'm hoping that we see an explosion of boutiques, Right.Where people come in and create very highly curated e commerce experiences for certain kinds of game verticals or genres or art styles or even just people's individual taste.And I think enabling those stores within a larger platform might be a really interesting strategy for esteem.
Speaker B
Right.I almost think as you're saying this, my mind goes towards our friend Nathan Garriott, who was from Annapurna that then started Fictions.
The role of a publisher maybe becomes increasingly important in that we know that any game Nathan and his team are going to co sign and help put out actually matters and is important in the same way we look at an A24 film in this way.And so weirdly, I do think publishers might have a rise back in this current moment.
Speaker A
Absolutely.In fact, when I took over the third party business at Activision back in 95, 96, we used to say that we had a house style.
Speaker B
Right.
Speaker A
There was like a certain kind of game that we wanted to.We wanted our games to have a kind of consistency even though they were being sourced from a variety of third parties.And we didn't really have control over the development.We were picking things that were sort of an expression of taste.And I think that was interesting because people could look and say like, okay, you know, Activision third party is making kind of dark, gory, hyper violent games or whatever our style was at the time.Right.And I think that was again sort of that publisher's imprimatur.Really, really important.
Speaker B
Yeah.Maybe Humble Bundle comes back, maybe you go full circle and you have something like that come back on the boutique games side.
The final durable layer is liability, including age verification, compliance, rights frameworks, and insurance-like guarantees. They argue that these painful systems create long-term moats and valuable verified data.
Speaker A
So the last of the five is liability and corporate insurance.Right.And I think this has certain overlaps with trust, as you'll see.I think they're quite different.Different.
But the general version of this is sort of regulated industries compliance, liability bearing surfaces.Right.So in the game space would be age verification, COPPA compliance, gambling and loot box rules, the eu, AI Act, IP chain of title performer rights frameworks that we're going to see out of the SAG and after agreements, platform certifications.Right.Like these ideas that there is a.There is some entity that is providing a liability guarantee in this.Right.Some kind of corporate or insurance company or whatever it might be.And I think those things are really important and those are going to be even more important when the sort of usual cluster of kind of identities and corporate Persona are no longer really in play in an AI future.And so, so I think we're going to see age verification regimes we're going to see tighten all across the industry.
Last night Roblox reported their earnings for their Q1 earnings and they talked about how age verification had made an impact as we had predicted in our Roblox episode.And they got hammered, as we predicted in our Roblox episode.And at least at the market open this morning, we're trading down about 20%.So I think that, look, age verification is a really important aspect of this sort of liability and insurance regime.And I think in the future when we have the sort of DMCA for AI, we're going to have compliance.
Speaker B
Right.
Speaker A
But necessarily.And there's going to be opportunities for people to build competence, whether that's internal competence or whether that's outsourced competence from.From third parties like compliance outsourcers, identity verification platforms and even billing and payment companies, as you suggested earlier.I think Stripe will play a role in this.Right?
Speaker B
Yeah.
Speaker A
And I would also point to the.I tell the sort of Axolas and the other payment processing companies in the world to be.They should go to school on what Stripe is doing.Like Stripe is crushing it in terms of how they're preparing the company to exist in the this AI paradigm shift.
Speaker B
Yeah.I mean they continue to operate at a different level.Just true excellence out of them.I think there's a piece of this which maybe Roblox is the one example that we can point to of these regulatory things and These insurance things and all of this, they create headwinds, they're not easy to do, these are not an easy thing to implement in your company.
And.And there will be pain, I guess.There's no simple way.There's a reason why it's hard to do these things because they're almost acts of violence on your company.And the way that you've structured, because you're trying to put up safeguards, you're trying to look out for your users, but long term, not even just in an AI sense, they become just traditional moats, they become a thing.
Speaker A
Oh my God.Yeah.Look, without giving away too much, it's like in a certain company that a well known company that I work with on the board, they're also going through a period of age certification which is obviously going to be wrenching and disruptive on a certain level in terms of excluding certain individuals from the sites and making it more adding friction to the onboarding, et cetera, et cetera.But once that data is in and once you know that you have age verified users, or non age verified verified users for that matter, that context becomes its own source of value.
Speaker B
Right.
Speaker A
That becomes a really, really important asset that could be outsourced and federated and sold to others.And so going through that pain and suffering is going to actually come.You'll come out the other end of that with something very, very valuable.
They close by previewing the broader AI production stack and naming tools across creative and business workflows. The episode ends with advice to think in terms of flexible workflows rather than fixed tools, since the landscape is changing rapidly.
Speaker B
I could not agree more.
Speaker A
All right, we've gone long and we were intending to finish this episode by talking about the new AI production series stack, which I was going to just mention some applications that people should look at, but I think this almost deserves its own episode at some point in the future.
So I'm going to just throw a few of them out there for people to look at if they haven't looked at them.And then I think we wrap this up and we move on.Because I think we could do a whole other hour just going through.Not just the actual artistic production stack, right, the sort of coding and design and art, but also customer acquisition, community management, finance, monetization, governance, compliance.There are AI companies operating in all of these spaces.
And I have recently with another company I'm working with, developed essentially what you would do if you were starting a studio from scratch and you were intending to use the most AI you could possibly use.And I've put together essentially my own sort of idea of what that constellation of services, of tools, of companies, of all of those things would look like.And I guess on a certain level that probably has some competitive value so maybe disclosing it on this podcast isn't entirely smart, but man, the tooling has exploded in the last 12 months.And, and like again, if you did a drive by a year ago and you were like, oh yeah, mid journey, that's never going to be that useful to me in making a game.Go look at Meshi.Go look at what's happening with Blender.Like Blender is essentially turning itself into an AI programmable system.
Like it's crazy.You know, Runway in the linear media space.Adobe Firefly 11 Labs for voice.
And that doesn't even begin to touch on some of the sort of business enabling things that have been developed that are incredible.So we'll come back, we'll do another episode on this.Hopefully the thing that I'm working on will have launched by then so I can disclose some of this stuff a little more fully.But I think we've done enough for one day.
Speaker B
Yeah, I mean, look, I think again, the disclaimer on all of even that stuff is just, it's moving so fast.
So even if we did do a snapshot of here's the hundred tools we would recommend and it changes so fast, Mitchell literally wrote a primer that I read that is amazing and super in depth, but 40% of it might have already changed in what is the best practices.And so that's just again the disclaimer on even this side of it.But look, for someone who is curious and actually wants to immerse themselves into this world, there is infinite resources out there.And look, I'm also, also happy to be a resource in that I think I try my best to spotlight these things on social media and others.But yeah, there's infinite ways to go and learn and try out these tools right now.
Speaker A
And I guess the advice that I would give what I've learned over the last six months or so, in particular in the last three months or so, is that the way to approach this is from a layer of abstraction above the tools, right at the layer of abstraction where workflows live, essentially.And to really understand those workflows, because the tools underneath them may need to be substituted as certain frontier models excel at certain things.And certain of them, they're going to start differentiating, they're going to start specializing.We already see it happening.Google's tools, which we've not really talked that much about, Gemini and nanobanana and things like that, are getting way better.
And I know certain, certain people prefer them for certain kinds of workflows and for certain kinds of tasks and so I think like aligning yourself with a frontier model and saying I'm a Claude person and I'm just going to do everything in Claude is stupid.
Speaker B
Right.
Speaker A
It's like you have to build workflows that have the flexibility to substitute these new tools because the tools world is so dynamic that it's really impossible right now to predict what's going to be, be what's going to be a leading edge indicator a year from now.
Speaker B
Right.
Speaker A
And so you really don't want to make massive investments in the kinds of workflow, you know, internal workflow processes that are going to be rigid and won't necessarily be flexible enough to accommodate the fact that what's happening under the surface right now is just insanely dynamic, like at a level I've never seen before in the, in, in the 40 years that I, that I've been working in, in software.
Speaker B
Oh man, we are, we are definitely in exciting times.And yeah, than.Thank you so much for putting this episode together.
Speaker A
You too.And we'll be back next week.
Thank you everyone.
There.