Epic CEO Sweeney：AI将重塑游戏开发，效率提升或达10倍

Epic Games CEO Tim Sweeney believes artificial intelligence has crossed the line.

In an exclusive interview with GamesBeat, Sweeney said he believes AI is transforming game development, one that could make studios dramatically more productive, lower the risks of experimentation and unleash a new wave of game genres.

But he also sees a growing divide between developers embracing the technology and those resisting it, warning that studios that fail to adopt AI could find themselves unable to compete. And, as he tweeted recently, he believes rival platform Valve isn’t helping matters. This kind of conflict over the usefulness of AI is playing out across gaming, as yesterday’s survey of 2,600 game professionals from GDC Festival of Gaming showed.

In our interview that also included Marcus Wassmer, Epic’s executive vice president of engineering, Sweeney described how rapidly improving AI tools are already changing the way Epic’s developers work. Sweeney estimates that AI is delivering productivity gains of four times or more in many areas, with improvements potentially reaching 10 times as the technology advances. Wassmer cited an early Unreal Engine web deployment experiment that achieved as much as a 20-fold productivity improvement in its prototype phase.

The executives’ comments offer a glimpse of how Epic sees AI reshaping not just the production of games, but also the economics and creative possibilities of the entire industry. And that matters because their Unreal Engine is a foundation for developers to participate in that gaming economy. Sweeney argues that AI will allow smaller teams to accomplish work that once required much larger staffs, while making it far less expensive to prototype, test and discard ideas that don’t work.

Our conversation touched on the controversy around overhyped innovations that have swept through gaming, such as blockchain and the metaverse. But Sweeney argued AI is different.

He thinks it’s coming just in time to address one of the biggest problems facing triple-A development: the enormous cost and risk of making games that don’t turn out to be fun enough. Sweeney said major projects have sometimes brought their playable experiences together too late in development, leaving teams insufficient time to experiment and improve them. With AI accelerating prototyping, developers could find the fun much earlier and explore many more possibilities before committing to a costly production.

“We’re back in an age of experimentation,” Sweeney said. He predicted that faster development could lead to new genres and fresh combinations of existing gameplay ideas, particularly in areas such as shooters and massively multiplayer online games, where large-scale experimentation has been limited by the cost of development.

Wassmer described how Epic developers are already using AI to make changes to games during playtests, generating a new build while a session is underway and then resuming with the requested changes. He also discussed how AI is helping engineers and designers tackle projects that might previously have taken years, sometimes compressing the initial work into days or weeks.

The company’s recent Unreal Engine 5.9 preview, which adds a web GPU target, provided another example. Wassmer said a small group of engineers used fleets of AI agents to accelerate the work and get a preview ready for Epic’s Mega Jam this week. The prototype came together in a fraction of the time that such an effort would traditionally have required. Epic is now cleaning up the implementation before calling it a finished release.

Sweeney criticized Valve’s requirement that developers disclose certain uses of generative AI on Steam, arguing that such disclosures have become a stigma that can overshadow the actual quality of a game. Good and bad games can be made with or without AI, he said, and consumers should judge games on their merits rather than the tools used to create them. He said user game reviews are a great way for fans to signal any problems related to AI quality, or to tell players if a game is AI slop.

He sees AI adoption as an economic necessity, not simply a matter of personal preference. Developers in Asia, he said, are moving quickly to incorporate the technology, and studios elsewhere risk falling behind if they refuse to do the same. Sweeney compared the transition to the move from horses to cars and from assembly language to higher-level programming languages, arguing that technological progress will continue despite political opposition, market volatility and concerns about overhyped AI companies.

Wassmer emphasized that the latest advances have made AI substantially more useful than it was even a year ago, particularly in coding and 3D content creation. He pointed to the growing capabilities of new models and said developers who dismissed the technology previously should take another look.

For Epic, the argument is that AI is no longer a distant possibility or an optional productivity tool. It is becoming a defining technology for how games will be imagined, built, tested and delivered.

We last caught up with Sweeney after Epic scored antitrust wins against the mobile platform owners. Here’s an edited transcript of our interview with Sweeney and Marcus Wassmer, executive vice president at Epic Games. (Note: We interviewed Sweeney before Nvidia’s Jensen Huang resigned over alleged affairs that raised corporate governance issues, and Sweeney has shared his opinion about that on Twitter as well).

GamesBeat: I noticed that you’ve gotten very active on Twitter. You moved into some strong stances. What’s been motivating you? Maybe starting with the criticism of Valve for requiring disclosure that AI was used to make games.

Tim Sweeney: Let’s back up. The reason I’ve been a vocal advocate of AI–we’re seeing absolutely massive gains in all aspects of game production. Code, content, workflow, testing, bug fixing. It’s all accelerating by industry-changing multiples, often 4X productivity gains. More often than not.

I’ve been traveling around the world. We had Unreal Fest in Shanghai, Seoul, and Tokyo in the last few months. Every developer in Asia is adopting AI as rapidly as possible and using it for maximum acceleration. It’s really clear to me that large-scale game developers–setting aside indies, where there’s always going to be specialties and niches, large-scale game development will be totally changed by AI. Teams that don’t adopt it will absolutely be left behind and rendered uneconomical, non-viable compared to their peers.

Even if they were to legislate, in America, that you can’t use AI, for example, Korea, China, Japan, southeast Asia, South America, everyone is going to be adopting it. When one team can make a game four times more efficiently – 10 times more efficiently soon, perhaps – than another team can make the same game, there’s just no possibility of competing the old way. No game developer has 75% to 95% profit margins that they can put at risk by not being as productive as possible.

Because there’s a big resistance movement against AI that seems concentrated on trying to bully developers into not adopting it by attacking individuals, by review-bombing projects and so on, I’ve been trying to make my views known to as many game developers as possible. We have to power through this and build awesome games, use maximum acceleration, because it is a survival issue for all the companies in the industry. Where there’s a barrier to adoption of AI, I’m trying to advocate both by speaking, but more important, by Epic acting and doing awesome stuff. Delivering new engine support for AI features. Sharing our best practices with developers, which we’ve done now many times with videos and talks and presentations about how to use it. That’s the overall context.

GamesBeat: I agree with a lot of that, but for the sake of fleshing out the discussion here, I’ll play a bit of devil’s advocate. One of the bigger points on the other side is that this is not necessarily a political stance by Valve, but someplace where they’ve decided that consumer transparency is important. They have concerns about AI and want to guarantee quality. Perhaps there is a lot of reason to make generative AI use a disclosure requirement. On that point, whether consumers really need to have this information, what do you think about that part of the argument?

Sweeney: I don’t buy it in the slightest. Steam already has a review system that’s very much intended to capture all facets of a game’s quality, scope, entertainment value and so on. That should be completely independent of decisions about particular tool usage. You can build good games and bad games without AI. You can build good games and bad games with AI. You can repeat those distinctions for all engines, all development tools, all methodologies, all countries of origin, all teams and so on. If we were to identify AI as associated with bad games, that’s just unfair to developers who use AI to make good games. It’s a guilt by association pattern that all developers need to push back on.

In particular it’s being used toxically here, because it’s absolutely being used by anti-AI advocates to undermine the reputations of developers and the viability of their projects. The existence of a disclosure has become the primary news story for the launch of several new games, rather than the game itself. That’s a new trend. A game is announced and it comes up with a pre-order page on Steam. That used to drive interest and enthusiasm. People would comment on the actual merits of the game. Does it look good? Does it look fun? Is this game for me? But the AI disclosure has become a scarlet letter, which drives attacks on games not because of their quality, but simply because of the choices of how they were developed.

This is absolutely a political issue. Major politicians have run campaigns based on anti-AI sentiment, anti-data center sentiment and so on. It had a major effect on the U.S. election cycle this year. There’s no question that this is a political issue, and a partisan political issue.

GamesBeat: Do you think it’s the kind of AI that’s the problem? They aren’t requiring a disclosure for vibe coding, for programming, but they are for generative AI used on the creative side.

Sweeney: It doesn’t make any sense to me. If one has a moral argument about AI, I see no genuine distinction between code, content, and the many other things AI knows and understands. AI acceleration of testing, AI bug fixing, AI analysis of software development practices, whatever one’s argument for or against AI, it should be roughly content neutral. Programmers and artists are people who make work that’s to some extent creative and to some extent technical.

GamesBeat: There’s a different kind of tradition here in the west of suspecting every new thing of being a scam. That started with free-to-play games and the associated monetization methods. We had it for blockchain. There are people who just throw this baby out with that bathwater and say AI is no different. It’s just one more scam. The house of cards around AI is going to fall because it’s way overvalued and there are consequences to that. Given that that’s the western tradition, and all those things haven’t been greeted the same way in Asia, how do you convince people that this is different? That this is a genuine advance for everyone.

Sweeney: Every wave of technology has had a scammy underworld to it. Even the most legitimate trends have people picking up on it with ill intent or poor practices. AI is certainly no exception. But the critical thing is to separate actual good products and solutions from bad ones. Cryptocurrency brought some improvements to the treatment of digital assets. There were also numerous NFT scams where items were sold that ultimately had no value or utility and people were ripped off. We need to separate the good from the bad. One could support some AI companies and not others. Treating an entire field as bad because of a bad example in it is not going to give us good decision-making.

Marcus Wassmer: If I could add to that a bit, it’s just a straight-up difference in kind between those things, to me. NFTs, NFT scams, even the notional uses of that technology are relatively narrow. The proof of this thing being useful is in the actual utility. If you look inside of Epic, we have a channel where people share their experiments, which is really cool. Across the board, people do all kinds of wild stuff. We get people making cool custom games to play with their kids. They sit down with their five- and six-year-olds and have them ideate things they want to see. They take videos of it. It’s super fun.

We have UX designers who used to only work in mockups directly committing UMG to the game, which is a night and day superpower for them. We have engineers who are able to sit and iterate–the AI doesn’t necessarily one-shot the code for them, but they can sit and iterate, and we’ve had a few projects that I know for a fact would have taken five or six years to complete, they get done to good quality in one week. It’s empowering for a lot of people. The across-the-board utility is a distinguishing factor here.

GamesBeat: This is where I swing over to being in favor of it as opposed to being a devil’s advocate here. It feels like, in the last few months, things have changed a lot. So much that people need to take another look. If you made up your mind about AI two years ago, in the last few months we’ve seen things like preventing cybersecurity attacks. Becoming aware of vulnerabilities and patching them with the same tools. It’s starting to become indispensable. Things have emerged more recently as far as the quality of what you can do.

Wassmer: That’s right. Each new model is significantly more capable than the last. Claude Code was only released some time earlier this year. Certainly internally we’ve seen some big turning points, like when Astra came out. That turned a corner on 3D content generation and spatial understanding. It was quite good. Opus 5.5, which came out recently, we found has been really good for code. Things have gotten better since then. There have been turning points where the general utility has gone up with a step change. People should take a second look. If you tried it out last year and thought it was no good, you should absolutely take a second look.

GamesBeat: Tim, you were also more vocal about the need to use AI across Unreal Engine and so on. Did the timing of the arrival of some of these new, more interesting examples of the technology influence you to speak up more?

Sweeney: The big thing that changed is AI became really good at writing code and assisting with 3D scene creation in the past nine months. Having tried a lot of this in the early to middle part of last year, it wasn’t nearly as gainful. One could imagine last year that some companies would use it, some wouldn’t, and it might not make much of a difference. But as of this year it’s so clear that the utility is massive. The productivity gains are so huge that any company that doesn’t adopt it is going to face real existential peril.

GamesBeat: Do you still have some worry? In some ways–you mentioned the election. There’s been the crash of AI stocks, and this fall for Jensen Huang as well. Is the perception of AI getting worse now? People are seeing some negative consequences coming. However good the technology is, politically or commercially things are not necessarily looking like they’ll go the same way as in Asia.

Sweeney: I think this is like transportation evolving from horses to cars, or the move from assembly language in the early days of programming to C and C++. This is an inevitable change in how technology works. We’ve seen, in each of these technology changes, ups and downs. There was dot-com excess and a bubble that burst, but then look at what came out of that successfully. Amazon, Google, and many others ultimately emerged to make massive gainful uses of the internet.

We’ve seen some excess here, and that’s reflected in the stock market, as well as the stories that different AI companies are telling. We’re now seeing some backlash to overpromising and overhype. We’ll get back to a level of normalcy. But the 4X productivity gain is not something you can ignore.

Wassmer: The utility is more durable than whether people like Jensen–it really comes down to the fact that no one has a monopoly on the technology, with all the open weight models out there. In the end everyone can take their cards and self-host the stuff. There’s no putting the genie back in the bottle.

GamesBeat: You have UE 5.9 coming. There’s an interesting aspect as far as some very strategic aims you have with Unreal making it to the web. How important is that particular advance? Do you see that as something that will stretch out over time, or will there be real benefits in the near term?

Wassmer: That’s an interesting one, 5.9. I think it did a couple of things for us. The most obvious, yes, we wanted to get the web target out ahead of the Mega Jam this week, so people could use it. Obviously we see all the cool stuff people are doing with Three.js and other things. Why not Unreal Engine? We should be in the conversation there too. That worked out pretty well.

This preview was done at lightning speed. We’ve been pushing changes into UE 5.9 for various reasons for a few months, but I think it was only a week and a couple of days ago that we decided to just go for it. We had a couple of engineers managing a few fleets of agents to try to get the web GPU target out there. We learned quite a few things along the way. That version is a preview for a reason. We got it out by the deadline, but we have some cleanup to do. The team is busy cleaning up before we’ll call it a clean 5.9 release. Still a little work left to do there. But it was a nice existence proof for us of how fast we could go on a thing if we wanted to.

I just said to a lot of the team today, I think our intuitions for how long things ought to take are probably wrong now in most cases. That doesn’t mean everything can be done in a day, but some things that used to take three months or four or five, it’s not obvious–maybe you could do it in a week or two or three. We’re in a phase of–now we need to try. We need to be ambitious and try to learn, and also learn the mechanisms we need to go at speed while also maintaining code quality and engine quality. We’ll get there.

So yes, we are seeing gains and we’re learning a lot from that 5.9 experiment. We’re going to take that forward for faster engine development.

GamesBeat: I know some of the other folks working on this web problem. Alex St. Louis had a team of two working for five years trying to get this done. They have the SimplyStream technology (to take Unreal games to the web). Is this the point, though, where AI really matters in terms of being able to finally get this done? People have talked about it for a long time, but was it AI that pushed this over the finish line?

Wassmer: A little bit. Tim can keep me honest here, but at least speaking for myself, I think the difference with AI now is that it lowers the opportunity cost to try certain things to a low enough degree that you can just go for it. Previously this sort of thing would take many people many months to accomplish. That’s a big deal. You need to plan for that. You need to consider what else those people might do instead of that in the meantime. Now, if you can compress that timeline, the opportunity cost is still there, but it’s small enough that you can take bigger swings.

Sweeney: Marcus, would it be fair to say that it was something like a factor of 20 productivity increase than if those people had done it all by hand?

Wassmer: I think that’s fair for the initial prototype. We’ll see how that shakes out for the final version. It might be more like five to 10 in the end for the final version. But it’s significant, the number of man-hours.

Sweeney: It completely changes the calculus of what products to build and how to build them with those sorts of productivity gains. A team of 20 can now accomplish what a team of 100 might have previously. The answer won’t be to just build the same software with less people. It’s going to be building more features, more capabilities. Exploring more possibilities, recognizing that some will succeed and some will fail. Every game developer trying things can try a lot more ideas and be willing to throw some of them out, because it’s no longer so expensive to try. Every game will be improved quite dramatically.

Even if it only improves the speed of shipping software by a factor of five, it probably improves the productivity of trying a rough prototype of something by a much larger factor, 10 or 20. Try out a game idea. See if it works. If that can be done in an hour instead of a week, that completely changes the nature of game development.

I think we need to back up and acknowledge that one of the major reasons why a number of triple-A releases have failed is because they were not fun enough, and the reason for that was because the playable game had only come together so late in the project that that the team wasn’t able to iterate and improve it enough before they felt they had to ship. If we can try ideas much faster and find the fun in games much faster, we can greatly improve our success rate with high-budget games.

GamesBeat: It’s interesting to see Roblox move in the same way, kind of breaking out of their box. The strategic possibilities here–if you had been able to have this technology in your pocket five or 10 years ago, how different would all of the litigation have been with Apple and Google? You now have a viable alternative to existing platforms if you can go to the web.

Sweeney: Not on iOS, because Safari is so crippled. If you try running these web games on an iOS device, it’s horrible. It doesn’t support 64-bit, so you can’t write serious scaled programs. A lot of the features needed for games are disabled by default, so if a normal user goes to the website to try something it doesn’t work. Apple in particular has a defense in depth strategy to prevent competition from emerging on the web, starting with making Safari bad as a web browser engine, and continuing by blocking other web browser engines from existing. Apple bans competing web browser engines except in the U.K. and Europe, where they’re required to support them. In my opinion, it’s to ensure that the web remains bad for games, so they can force those apps to be native apps behind their 30% paywall. So it wouldn’t have helped with that.

Let me tell you about how Unreal Engine 1 development worked, when it was just me building up 250,000 lines of code in three and a half years. There were times in which I decided I needed to make a radical change to the codebase – moving from C to C++, moving from DOS to Windows, moving from all C++ code to a scripting language, UnrealScript, side by side with it, and then moving between literally seven different renderers. There were times when I decided I was going to make a massive change, and I spent the next two weeks, 18 hours a day, just typing code and refactoring the entire code base to a new way.

This became impossible in Unreal Engine starting with Unreal Engine 3. The codebase had grown to millions of lines of code. The cost of taking the entire team offline for what would be months in order to do a major refactor was completely prohibitive. There were entire categories of improvements that could have been made to the engine, but we had to forego them because the cost was way too high. We made some. We replaced the renderer completely with the Unreal Engine 4 transition and built massive new systems for Unreal Engine 5, but an awful lot of the engine is just what it’s been all along.

For the first time now, we’re back to the days in which a person, or a couple of people, can, over the course of a couple of weeks, massively change and progress the engine. It’s truly a new golden era of software development. As more companies realize and start to capitalize on this, we’re going to see breakthroughs that stupend the whole industry.

GamesBeat: Do you have some big thoughts on the future of games? Regardless of whether or not AI will enable it, what do you think about where we’re going to be in a few years?

Sweeney: One good sign is we’re back in an age of experimentation. The only games that could truly experiment with radical new game designs in the last few years have been small indie projects, because you just can’t afford to do that with a $30 million to $300 million project. There’s been a lack of experimentation in larger-scale games. But with these new productivity improvements, I think we’ll find a lot more attempts to find and create new genres, or major twists on games, because you can do it faster.

Many genres have been neglected. Let’s look at shooters. The two big breakthroughs in the past 20 years are battle royale and extraction shooters. In the first five years of shooters you had an explosion of possibilities. Then that slowed down because the games got too big and trying things became hard. Similarly, think of massively multiplayer online games. We basically have one template for those games that has been done at scale, the MMORPG, which is based on incredibly complex D&D-style role-playing game systems. We have a bunch of different takes on that, but it’s all largely the same formula.

With the ability of teams to build a project like that with perhaps one-tenth of the effort, or to at least experiment and prototype one, I think you’ll see a lot of exploration of new gameplay ideas, mixing and matching up old formulas in new ways, and new genres emerging that couldn’t have been found previously just because of the expense and the risk of the investment that would be required, and the unwillingness of companies to fund that beyond the indie scale.

GamesBeat: How can you get around the problems that VR ran into? We saw outstanding experimental technology from indies working on things we’d never seen before, but it never broke into the mainstream.

Wassmer: That was more of a hardware problem than anything. The hardware companies could never truly crack making it comfortable for most people to play those things for a long time.

GamesBeat: AI doesn’t have to worry about new hardware technology coming along. It can be applied across all platforms. But do you think some of the positives in the opportunities now might arrive when we get new consoles, new hardware? Or can it just happen now?

Wassmer: I don’t think there’s any reason to believe we need to wait for new consoles to see this. Like Tim said, you’re going to see experimentation. You can see it now. Here’s one of my little anecdotes. A lot of us at Epic have side projects where we play around. I have one with some friends. In the middle of one playtest for the game, we were taking notes, and not just taking notes, but giving instructions to the AI about stuff we wanted to change.

In the middle of the playtest we got a new build back. We had a save/load system, so just save where we were, shut the old build down, bring the new build up, continue the playtest with the changes we wanted, literally in the middle of the playtest. That’s the kind of iteration in development that people are going to get. It’s something a small team can do. I think the expectation of an explosion of experiments in gameplay is a reality. It’s going to happen. You don’t need a new console to see that.

Will it make it easier to more quickly–how do I want to say this? Every generation of console hardware, you see an evolution from launch titles to the big marquee titles at the end of the generation. If you go back and look, developers really figured out how to scrape every last cycle of the hardware over that period of time. I think what AI will probably do on that front is shrink that time frame for really extracting the most you can from the hardware. But that’s going to be true for existing hardware and platforms as much as new ones.

GamesBeat: We’ve talked a lot about the different constituencies here – developers, consumers, technologists. What about creators? You have a lot of those around Fortnite. How do you think they can be convinced to come around and put their voice behind new technologies and help pave the way for more acceptance?

Sweeney: This is their time. My impression is that most Fortnite creators are using AI to accelerate development. Roblox put out their announcement that 80% of Roblox developers are heavily using AI. The interesting thing about building a game that way, within an ecosystem like Fortnite or Roblox, is the development cycles are much faster. A week or two from idea to launch. Then often improvements daily, or multiple times a day, based on watching community feedback and trying new ideas.

Those are the teams that will benefit the most from this. Every week they’re going to be adopting the latest best practices and using them to make their development more efficient. They’ll be pushing each other and spreading best practices around. It will be immediately apparent when you play a game what breakthroughs were made. It will be a much slower process in triple-A development, where you have a three-year development cycle. In year two AI comes along and the team starts using it, but it’s going to be a fairly incremental adoption process there, rather than a radical reinvention of development cycles.

There’s a real chance of ecosystems pulling ahead of triple-A in terms of their capabilities available to creators and their adoption. You might see an actual inversion of leading innovation occurring in these much faster-turnaround projects.

Wassmer: I don’t think our developers need to be convinced. The developers that are successful on UEFN and other platforms are small teams that get after it. They’re hungry. I think they’ll already find every way they can optimize and go faster.

GamesBeat: I wanted to ask about a potential side effect that fast development has enabled, which is fast copying. It’s at a point where some folks are afraid to release a playtest publicly, because someone could steal the idea and get it out before the original creator. How do we solve something like that?

Sweeney: I think it’s going to change the way that people build, especially on small projects. You’ll see a lot more teams just launch and be playtested in public. They’ll build up a userbase as they’re finding the fun. Instead of a large public playtest, you’ll see a development team and a small number of handpicked friends building and playtesting a game, and then they’ll just launch. No marketing, the game just comes out. That brings back memories of the old days. In the 1990s a lot of games just launched, rather than having a long hype cycle precede them. That might be the way in the future.

Game developers have a way of routing around obstacles in the industry. We went through one in the early 2000s when Bittorrent came out and made piracy a whole lot easier. Back in the days when everything was at retail that was a real problem. What happened? Steam came online. Publishers like Riot and Epic started publishing games directly to consumers on the web, making it convenient to get in and download. Free-to-play came along and eliminated the reasons to pirate a game entirely, because you could get a game for free and spend money as you chose.

These latest challenges – copyright infringement, rapid cloning, rapid adoption of ideas from competing games – are going to be new trends that shape the industry, and we’re all going to have to adapt to that. This is the ground changing underneath us. There are going to be massive benefits, but also some drawbacks. We can’t pick and choose. This change is happening now. It’s permanent. We’re all going to have to do our best to adapt to it.

GamesBeat: Hollywood is also embracing the technology in some unexpected ways. Edward Saatchi’s company Fable is releasing this AI-made movie in theaters on December 4. They’re only using human actors for the voices. Similarly, Shortical and other companies are making short films with AI on social media. They’re getting episodes of these shows out in a day. They’re racing ahead of the point where AI enhances human creators and developers to a point where AI is doing the work. Do you think the game industry is also going to see pressure coming from very quickly made AI-only games?

Sweeney: I think so. Creatives love creative tools. It’s unfortunate that the leading AI companies introduced AI in such a brash and in many ways offensive way, but the tools are awesome. I think everyone is going to be drawn to them in their own ways as they realize what cool things they can accomplish. More and more people will ship awesome things made faster and better, games that couldn’t have been made anywhere near the same budget before. You’ll see an entire new generation of creators come out empowered by this. I think people will eventually come to love the change that’s occurring.

Right now we’re in an economy where everything is either a super huge budget production or a short-form video on YouTube or TikTok. We’re going to find that there’s a large design space in between that people can fill out. There’s a lot of awesome content that can be built without a high-end budget or production values for an audience that just loves it.

The post Epic Games CEO Tim Sweeney predicts AI revolution in gaming with 10X productivity and entirely new genres — if devs get behind it | exclusive appeared first on GamesBeat.