"A perpetually self-feeding meat grinder:" The futile search for developers who adore the latest prompt-gobblers from Unity and Google

"Prompt it. Perfect it. Publish it." That's the six-word sell behind Unity Spark, a new generative AI platform that will purportedly allow anybody to "direct" their dream video game. In a trailer deployed by Unity, Spark is expertly wielded to bless the universe with "corridor shooter set in space."

Unity Spark is being deveoped and promoted in strategic partnership with Google, which last week debuted its own experimental, prompt-munching game creation platform Playground. Much like Spark, Playground compels would-be makers to "create custom games in minutes using simple text prompts, no coding experience required."

Google claims users will be able to create custom games in minutes. Mercifully, it also seems to understand the lane it's currently occupying by insisting it's only targeting "hobbyists" or people "just curious about AI."

Unity seems more inclined to have its cake and eat it.

In a effusive blog post entitled 'Why we built Unity Spark,' Unity president and CEO Matt Bromberg stressed the technology isn't about "one-shotting" production beats, but rather using new technology help more people to express themselves.

"Asking AI to regurgitate an already invented game does not make you the next Hideo Kojima, any more than re-typing the first paragraph of War and Peace in your word processor makes you Tolstoy," he wrote. 

You know, he's right. If the post had ended there—a meagre two paragraphs in—we might have been on to a winner. Alas, Bromberg goes on to state that Unity Spark is for "the gamers who have spent a thousand hours inside a genre and know exactly what’s missing," which, to me at least, seems to indicate that he does want to make users feel a teeny bit like the next Hideo Kojima. 

He also notes that Unity Spark connects directly to the Unity Asset Store, allowing users to "build with thousands of high-quality, artist-created assets made by our community." Unity has confirmed that Asset Store publishers will be opted into supporting Spark by default. You have to wonder how they might feel about that? Anyway, back to Bromberg's big pitch. 

"Will some [Spark users] graduate to become professionals in our industry? Will some current professionals use Unity Spark as well to explore new ideas? We certainly hope both are true. We’ll see. One thing is certain: Trying to build a real game for yourself will only deepen the respect you have for the people who apply the art and craft of gamemaking at the highest level," he continues. 

"Making even rudimentary interactive entertainment is very hard, demanding expertise in many crafts—narrative, design, art, sound, UI, level design, front-and-back end development, systems design, scripting, just to name a few. Producing a polished experience of your own has long been beyond the capabilities of most people."

Almost every developer I have ever spoken too would agree with him there. Video game development is hard. It is painful and exhausting and finicky—and yet, everyone who reached out to share their views on Spark and Playground had nothing even remotely kind to say about either platform.

Developers feel platforms like Unity Spark and Google Playground directly undermine the act of creation

"From looking at Unity's demo, it mostly seems like a collaborative tool to have Unity pick out prefabbed assets for you so you can avoid having to place them yourself. Basically, doing none of the creative work that one might do when using prefabbed assets," says Pedalboard Games studio head Colin McInerney.

"One of the most interesting parts about game development (at least in 3D, avatar-driven games like the one showcased in the demo) is the level design—what goes where, and why. The message the developer tries to send to the player to engage them. Unity's basically promising to offload that for an underwhelming demo that looks worse than any of their previous demo projects that've shipped with numbered engine versions."

McInerney says he keeps trying to find reasons to not constantly dismiss generative AI production tools, but feels the act of creation is defined by a unshakable desire to realise a ludicrously specific vision. Or, to put it another way, it is the antithesis of what platforms like Spark and Playground are selling.

"I spent the bulk of the last two days adjusting a shader for my game's sky because I wanted it to 100% perfectly match the vision in my head, and it's still only 80% there. I don't know that it'll ever reach 100%, and if I do it'll because someone with better tech art skills and the ability to get that abstraction out of me puts in a monumental effort. This tool isn't ever going to be helpful to me," he adds. 

Then there's the problem of visibility and distribution. Game UI Database founder Edd Coates explains the idea of "democratising" production skills sounds wonderful on paper, but feels the reality will simply be a deluge of shovelware devs, cloners, and opportunists looking to flood the market in the hopes of earning a quick buck—drowning out those who actually care about the craft.

"[Even] if you do have a great idea that somehow gains traction? There's absolutely nothing stopping others from using that same technology to copy your game and release it themselves. It's a nightmare," says Coates. "So instead of making game development accessible to everyone, this is actually having the opposite effect and preventing unestablished developers from achieving any sort of success—regardless of the tools they've used."

"The accountability for what you've made is lost—because the cost to you has become functionally zero and the mistakes are the mistakes of the machine, not your own actions."

One problem that feels unique to generative AI tools—irrespective of where they're used—is how they demand a mind-eroding sacrifice. Knowledge for speed. Intent for output. Nuance for, well, whatever you'd call "corridor shooter set in space."

Biscuit Factory Games founder Dan Bergin-Holly worries that taking a shortcut will lead to a dearth of understanding within the video game industry."When iteration and creation costs you real time or effort you have to engage with what you're making in a different level. You have to ask why one feature will improve the game and if it's worth the time—and then when you're eyeballs deep in it at 2am trying to solve the one bit thats broken you learn that cost in a visceral and embodied way," he says.

"Similarly when the process of testing and debugging forces you to turn over the innards of a game you made over and over again, you are forced to reckon with your own thought process. A past version of you is reflected back at you in the terrible code you made, or the artwork you now hate to look at, or that jump feature you thought was good but actually hated making. If all that process gets abdicated to a machine for one-shot prompts, that introspection and consideration is lost. The accountability for what you've made is lost—because the cost to you has become functionally zero and the mistakes are the mistakes of the machine, not your own actions."

Over on Bluesky, Compound Fracture lead developer Austin Schaeffer offered a rather blunt appraisal of Unity's decision to continue doubling down on generative AI. 

"For an engine that has easily allowed us to do so much of our game DIY with a completely in-house render pipeline and tons of in-house editor tools, it's insanely frustrating to see Unity pushing things so hard in the exact opposite direction," he wrote.

Schaeffer notes he has been a devout Unity user since 2008, when he saved up money during his high school years to nab a Unity Pro educational license for $750 back when it was needed form post-screening and real-time shadows. To say he is disappointed in the company would be an understatement of some magnitude.

"The engine is unmatched for me and I hate seeing it die at the hands of such a foolish company," he continued. "[...] I'd like to go on record and say fuck GenAI. Fuck the companies that have made billions developing automated plagiarism machines and fuck anyone defending it. They don't make tools, they use the Earth's resources to shove human culture into a perpetually self-feeding meat grinder."

I'm sure you've reading to room at this stage, so lets close on what is easily the most succinct quote sent our way, courtesy of Super Spline Studios co-founder Aron Durkin: 

"If you can't be arsed to make it, why should I be arsed to play it, let alone pay you for it."