Skip to main content
If you click on a link and make a purchase we may receive a small commission. Read our editorial policy.

Is chasing artificial intelligence genuinely senseless? | This Week in Business

How can Roblox take a cut of sales that aren't happening if AI undermines the creator economy?

This Week in Business is our weekly recap column, a collection of stats and quotes from recent stories presented with a dash of opinion (sometimes more than a dash) and intended to shed light on various trends. Check every Friday for a new entry.

With a name like "This Week in Business," the expectation is that we talk about things that happened this week. Things in business, even.

But sometimes expectations go unfilled. And sometimes we plan to write about one thing, but then Unity shoves its foot into its mouth clear up to the knee and we feel obligated to point at the company trying to choke down its own femur and say, "I don't know, seems like a bad idea?"

So anyway, we're playing catch-up this week and looking back to two weeks ago at an interview we did with Roblox's chief technology officer Daniel Sturman about the company's plans for generative AI.

QUOTE | "It can act as an incredible on-ramp to creators on the platform, but it also can work in a way that can really accelerate existing creators. It's not a separate thing from Studio. It's all part of the creator tools we've been shipping for years; it's just accelerating them with the power of generative AI." – Sturman talks about Roblox's AI Assistant, a conversational tool that users can ask to pull assets into a project and code behavior for objects in a game.

Right now the AI Assistant pulls assets from Roblox's Creator Marketplace, but eventually the plan is to roll it out with 3D model creation, so Roblox developers will be able to ask for a model of a car and the AI will generate one that can then be tweaked on command.

Now I have what I think is a healthy skepticism toward AI. I question whether it will ever work the way backers are promising it will. I question whether the people rolling it out – those who swear up and down they want to do it in an ethical and proper way – will ultimately just shed any ethical concerns (and the people responsible for voicing them) as soon as it is financially prudent to do so, just like Amazon/Twitch, Microsoft, and Twitter have done.

I question whether it is even possible to pursue AI in an ethical way, given how much of it is essentially just plagiarizing the work of humans, but with an attempt to put a convincing veneer of originality over these thoughts the AI did not (indeed, cannot) have.

But let's assume for this column that ethical AI is possible, that it will work as advertised, and even that Roblox – a business frequently and convincingly criticized as exploiting children but one that still loses money hand over fist anyway – will be a responsible steward of this technology. Let's assume it "democratizes game development," which is a fancy way of saying "enables everyone to make games if they want."

It is an objectively Good Thing to have more accessible tools that make it easier for creators to realize their ambitions and put what's in their head on the screen, not least of which because it increases the diversity of the talent pool. A diverse talent pool also broadens the addressable audiences for games; it's no coincidence that the industry has shifted away from making games exclusively for young men as the people making games have shifted away from being exclusively young men. So bringing in more developers isn't quite a zero-sum game of more devs fighting for their slice of an identically sized pie.

Even in the best-case scenario, the democratization of game development is not a completely painless process

But even in the best-case scenario, the democratization of game development is not a completely painless process. We have seen repeatedly in this industry what tends to happen when the barriers to development are lowered. The mobile market gave developers easy access to developer tools and debug units, modest app developer fees, and digital distribution storefronts with minimal curation. As a result, we saw prices drop from $10 to $5 per game on feature phones while Apple's App Store launch lowered the bottom end of that range to $1, where it stayed for a moment before free-to-play became the standard and the conventional wisdom determined that premium pricing was practically a non-starter (barring outliers like Square Enix and Minecraft).

Over on PC, Valve built Steam into the default PC storefront for many. It gradually lowered the barriers on the platform until it stopped pretending to have standards entirely, and we were treated to years of "indiepocalypse" Discourse along the way. There are still loads of developers thriving on Steam and commanding respectable asking prices for their work, but discoverability on Steam has long been a puzzle for developers, and the "solution" to that puzzle is often deeply discounted games, which is bad enough when it works but I imagine particularly demoralizing when it doesn't.

As console storefronts have adopted more indie-friendly policies, we've seen similar changes play out even in the gardens that traditionally had the highest walls. And now we've got platform-holder-backed subscription services pushing games even further down the road from "creative output that people seek out and support through purchasing" to "undifferentiated slurry of content chosen for us by platform holders."

Now consider how this same trend would play out in Roblox, where creators make money – sorry, that should be "money" because they pay out in company scrip called Robux and actually cashing out is a costly hassle – by selling games and avatar accessories.

A Roblox screenshot of a car in a city at night. It is raining. A caption on the top left says "Make it rain," while text input at the top right of the screen shows a user asking for exponential rain with linear splash and circular mist and linear particle emitter tweens, each with their own interval time. The bottom right of the screen shows code generated by the Roblox AI tool based on those text prompts.
Paradoxically, AI tools will make it both easier and harder for developers to "make it rain"

Roblox is a free-to-play app, so the barriers to development at this point are essentially just the difficulty of making things. The barriers to making money off development on Roblox are considerably higher, however. And if generative AI tools lower the former barriers without affecting the latter, I would expect the platform to suffer similar effects to what we've seen on mobile, console, and PC storefronts, with plenty of pressure to drive down the prices creators are able to charge for their work. And that's keeping in mind that it's already pretty hard for anyone to stand out on the platform as is.

STAT | Over 15 million – As reported in the company's last annual report, the number of active Roblox experiences. And I call them experiences, because as we learned during the Epic-Apple lawsuit, the things people play in Roblox are not games because they are maps.

Roblox, of course, has a solution for discoverability: letting creators spend Robux in auctions to bid for featured placement in the storefront. Coincidentally, that solution also helps Roblox make money while raising the barriers to developers making money.

And this is what happens when Roblox is still willing to lose money quarter-in and quarter-out.

QUOTE | "Since our investment decisions are generally based on levels of bookings, we expect to continue to report net losses for the foreseeable future even as we anticipate generating net cash provided by operating activities." - Roblox CEO David Baszucki in the company's last quarterly earnings report.

Generative AI tools may make Roblox a more attractive place for some creators, but the same AI tools will make it a less attractive place to build a business

Imagine what happens when the company finally decides it's time to turn a profit. Unity is another company riding on nearly two decades of growth fueled by net losses, and we saw last week what happens when an outfit like that wants to turn the corner.

Generative AI tools may make Roblox a more attractive place for creators who aren't concerned with any kind of ownership or control of their work, but the same AI tools (and the company's general trajectory) will make it a less attractive place to build a business.

We asked the Roblox CTO about AI tools devaluing the work of the creators the company relies on, and he had an answer ready.

QUOTE | "What we've seen over the course of technology in general is a move away from the need to do things by brute force, and instead increasingly what will determine what a successful creator is will be that spark of genius they have on the idea or concept of what they want to go do. And we see that on Roblox already as being what separates a successful experience from a less successful experience, and we think that will just accelerate it." – Sturman suggests AI tools will benefit the genius developers who most deserve it.

We've actually heard this idea before in a variety of ways. It's basically the myth of meritocracy. The "spark of genius" will separate the wheat from the chaff. Best idea wins.

The problem with that is for as long as there's been a games industry, there have been sparks of genius utterly smothered by fast followers and clones.

Take the Magnavox Odyssey, for example. The pioneering console of the early 1970s had a table tennis game that will instantly look familiar to anyone who knows its more popular and successful clone, the Atari classic Pong.

STAT | $700,000 – The amount of money Atari paid to license Magnavox patents as part of a settlement over the blatant theft.

Screenshot of Atari's Pong, with a ball streaking toward a paddle on the left side of the screen. The score is 6-9.
Innovation, iteration, or immitation?

If the spark of genius is what matters, then why did 2048 become a global phenomenon over Threes? Why was Zynga able to build its empire on FarmVille and Mafia Wars while Farm Town and Mob Wars were footnotes in their success? Why is Fortnite so much bigger than the DayZ mods that developed the battle royale concept?

Well, you might say that the true genius in these cases wasn't the core gameplay innovations but the marketing. Or maybe it was the timing, or the business model, the pre-launch effort to build out a mailing list, the game's "streamability", the decision to be exclusive to one platform, the decision to arrive on every platform at once, or some secret brilliance that simply isn't evident to those of us watching from the outside. In Fortnite's case, you might say the genius was in bringing a wealth of disparate brands together in one experience, to which I would suggest that Epic owes Gamemaster Anthony a sizeable royalty payment.

All of this is basically survivorship bias at work, goalpost moving after the fact. The spark of genius always wins out, so therefore anything that wins out had the spark of genius and we just need to properly identify whatever it was.

People pushing AI want you to think there's some justice to who it will benefit and who it will harm, that the effects are deserved, in both directions

The problem with that is the spark of genius can also be luck. Or having a truckload of money. Or a particularly aggressive and competent legal team. Or access to a Scrooge McDuck vault of cherished intellectual property (like, for example, Scrooge McDuck and his vault). Or in the case of Facebook-era Zynga, a pathological lack of shame. These are all possible paths to success, but I would be reluctant to call any of them genius.

But that's not what Sturman wants you to think of when he talks about a "spark of genius." People pushing AI want you to think there's some justice to who it will benefit and who it will harm, that the effects are deserved, in both directions. Because it absolutely will harm people when their work is stolen and sold to make others rich, and that theft can happen either with the training data fed into the AI, or the knock-offs and clones developers will make using the tools Roblox creates.

And as for who it will benefit, I'm not even sure Roblox will ultimately be on that list.

Roblox makes money because creators use Roblox to make money. If AI tools devalue what those creators produce, there will be less incentive for creators to create on Roblox.

Who is going to pay for a bootleg Batman cowl or a blue cowboy hat when they can just as easily have the AI make them one if the only difference is typing the description into an AI prompt field instead of a search bar? It may not be quite as nice an item, but most people will take "free-but-wonky" over "good-but-costly" any day of the week. (And I'm just assuming the AI tools will be free; it will certainly be fun if Roblox gets people to rely on them and then starts imposing a per-prompt charge...)

Roblox actually seems to understand it may be hurting the opportunities to build a business, as it has announced changes to let Creator Marketplace developers keep 100% of their sales minus taxes and processing fees, and is converting that part of the storefront to using actual currency instead of Robux.

I see it as an olive branch of sorts to creators, an apology for knocking their gravy train off the rails, no matter how little gravy it actually contained. It also suggests to me that Roblox actually believes what it's telling people about the potential for generative AI tools in game development.

Unity's in a similar boat I think, as we talked about last week. The company is radically changing its monetization in a way it knows will be deeply unpopular with its users in part because of how it sees AI impacting the industry. It sees a future where its current model – largely driven by seat licenses developers buy for each employee working on a project – is hurt in a market where games are made by fewer developers, so this seems at least partly like a pre-emptive adaptation to that. Unity knew full well it would take a reputational hit in the process.

QUOTE | "I don't think there's any version of this that would have gone down a whole lot differently than what happened. It is a massively transformational change to our business model." – Unity CEO John Riccitiello, in an all-hands meeting with employees this week.

So if these two companies believe generative AI would upend their current business, why are they among those leading the charge for AI in game development?

It seems a bit self-defeating, but also very much in keeping with tech and gaming startup behavior. Disruption is an inherent good, and whoever disrupts a field first is most likely to be at the top of the heap when things settle. So even if you're already on top of the heap, there's plenty of pressure to change with the times and not be seen as complacent.

As for how that disruption will impact creators, Roblox thinks it has it all under control.

QUOTE | "No one has an interest in a race-to-the-bottom sort of economy. That would not be what I would call a healthy economy, whether it was on Roblox or in the real world. That tends to discourage creation when you have that. I feel we have some really good systems in place, and economic research is something we continue to invest in and spend a lot of time on as well. I think as long as we build the right economic systems, it will come down to having the good idea versus the manual labor of building the item." – Sturman explains the company's comfort with the AI pursuit.

Game companies have been pledging to "build the right systems" for years to hand wave away seemingly intractable problems

Investing in economic research sounds great, but game companies have been pledging to "build the right systems" for years to hand wave away seemingly intractable problems with in-game economies, storefront discoverability, and toxic behavior. Some of those systems are better than others, but even under the best systems, all of these very hard problems remain very hard problems.

I'm not sure its possible to build systems that will successfully mitigate or moderate the firehose of content that would be produced by AI tools that worked as promised. Just look at the state of modern search engines and mobile app stores to see how feeble the attempts to deal with discoverability problems have been under the non-AI-powered status quo. Or look at any social media platform to see how automated moderation systems cannot hope to keep up with the amount of harassment and abuse these platforms enable. (And in many cases, bad actors can actually turn such systems into yet another tool of harassment.)

Like many other AI advocates, Sturman believes the same formerly bottled genie that visits such curses upon us will deliver us from them as well.

QUOTE | "AI will open some of what I'll call 'new frontiers' in how we think about safety, but it's also an incredible tool… Of course AI brings some unique challenges to how you do moderation, but it's also bringing some incredible opportunities to how you make moderation better and stronger." – Sturman says only AI can solve the safety and moderation problems that will be created by AI.

This resulting arms race would certainly benefit companies selling AI services – and other services to mitigate the impact of that first set of AI services – but for the rest of us? I'm skeptical.

If AI lives up to the hype, a platform like Roblox is almost superfluous

Now if AI lives up to the hype, a platform like Roblox is almost superfluous. The main reason for creators to put up with its Byzantine business model and industry-trailing revenue share is because it's easier than using the alternatives. If AI makes those alternatives every bit as easy, what is Roblox's value proposition?

But if generative AI tools are a complete flop for small-scale or hobbyist game development, then Roblox has wasted its investment in the tools and given up its share of Creator Marketplace revenues for nothing. That's not an existential threat by any means, but it's not great for the company, either.

Honestly, I think the best-case scenario for Roblox here is that the AI tools companies are currently pursuing fail to revolutionize game development in any meaningful way. They optimize some tasks and perhaps lower some technical barriers to entry, but they don't significantly decrease the commitment or time investment required to make a business out of selling games.

Such an outcome would be good for Roblox if the novelty of the tools can entice enough would-be developers to take a peek at the platform and try it out. Maybe some of them stick around as creators, maybe some find themselves there more as players.

It's not a particularly bold vision of the future, but it might be the best outcome for everyone in the long run.

Correction: The original version of this story mistakenly referred to Roblox's Marketplace (the storefront for avatar-specific customizations) as its Creator Marketplace (the storefront for assets and tools) in several areas. It has been amended.

The rest of the week in review

QUOTE | "This is a significant milestone for the merger and a testament to our solutions-oriented work with regulators." – Activision Blizzard CEO Bobby Kotick brags to staff about the UK CMA giving preliminary approval to the Microsoft acquisition.

QUOTE | "It really isn't at all about Sony or Microsoft's platform, it's really about the future of technology. [UK Prime Minister] Rishi Sunak has said they'd like to be the Silicon Valley of Europe or of the continent, and if deals like this can't get through, they're not going to be Silicon Valley, they'll be Death Valley." – Kotick, giving an example of solutions-oriented work/implied threats in a February interview right after the CMA initially blocked the deal.

And of course, if companies like Activision Blizzard took their business elsewhere, what would the UK do without all those taxes it doesn't even pay?

QUOTE | "We are listening, talking to our team members, community, customers, and partners, and will be making changes to the policy. We will share an update in a couple of days." – Unity, in a public apology post Sunday night. As of Friday morning, that update had not been shared.

QUOTE | "The trust is flat gone forever but I wouldn't mind if they ditched this whole plan and tried again for something more sensible." - Papers, Please and Return of the Obra Dinn developer Lucas Pope had plenty of company in telling us Unity's Runtime Fee policy has irrevocably damaged the company.

STAT | 500 – The number of studios that have pledged to turn off all IronSource SDK and Unity Ads monetisation in their games in protest of Unity's Runtime Fee plans.

QUOTE | "They believe our targets (Planned Parenthood and C.S. Mott Children's Hospital) would NOT count as 'valid charities' and more 'political groups'…" – Orgynizer developer LizardFactory says that even though Unity pledged to waive its install fees for games going to benefit charities, a Unity rep told the studio it would not be eligible.

It's honestly quite impressive how many people Unity has managed to anger and alienate in the past couple weeks. We may never see a run like this again unless Kotick (the undisputed anger and alienation GOAT) can dig deep and put together something special on his presumed way out after the Microsoft deal closes. We are witnessing history, people.

STAT | Everything – What we learned from the massive Xbox leak that basically laid out Microsoft's hardware and Game Pass strategy for years to come.

QUOTE | "We've seen the conversation around old emails and documents. It is hard to see our team's work shared in this way because so much has changed and there's so much to be excited about right now, and in the future." – Xbox head Phil Spencer assures us all the plans have evolved so we should pretend like we didn't know this stuff was happening when they eventually announce it and it's exactly what we thought it would be.

STAT | 36 – The number of employees laid off at Embracer studios Beamdog and Crystal Dynamics this week as the company's restructuring continues to reverberate through its sprawling studio system.

QUOTE | "One of the saddest things that happened during the period of loose money was that investors encouraged – sometimes even pushed – a lot of previously self-sufficient studios into expanding their team and budget too much, to the point where budget outpaced market potential and the studios were no longer sustainable without investor support. When money became tight again, investors pulled their support, and the studios had to downsize or even shut down. This damage could have been avoided with the more cautious approach people are starting to adopt now." – In a panel on the state of investment in games right now, Hooded Horse president and CFO Snow Rui says money is tight, but that may be for the best.

QUOTE | "It has to hurt more." – Rockstar co-founder Sam Houser, 10 Years Ago This Month, suggests game development should be like Apocalypse Now. This was years before the company's Red Dead Redemption 2 crunch scandal, but years after the company's Red Dead Redemption crunch scandal. (Also Grand Theft Auto 5 is a decade old now, which is wild.)

STAT | 25% - How much of Bandai Namco's revenue comes from sales outside Japan, as we were told by Bandai Namco Europe CEO Arnaud Muller. He said the company aims to eventually grow that figure to 50%.

QUOTE | "We realize we're total rookies when it comes to creating Souls-likes. This fact made us less concerned about the competition in the market and more concerned about delivering a game that gamers can really enjoy while meeting our own goals and expectations." – Game director Jiwon Choi talks to us about how Round8 Studio and Neowiz decided to explore unfamiliar territory with the just-released Lies of P.

STAT | 1 – The number of games Glen Schofield saw through to completion during his time at Striking Distance Studios, the studio he founded in 2019 and left this week. The game was The Callisto Protocol, which reportedly fell short of sales expectations. The studio laid off 32 employees last month.

Read this next

Brendan Sinclair avatar
Brendan Sinclair: Brendan joined GamesIndustry.biz in 2012. Based in Toronto, Ontario, he was previously senior news editor at GameSpot.
Related topics