Great episode of The Town. Matt Belloni interviews Edward Saatchi, the CEO of Fable. I found it fascinating, both for what it gets right and where I disagree. It also reinforces how challenging it is for everyone to build any sense of intuition around what’s possible, because AI continues to move so quickly.

Moving beyond tools

While Hollywood execs and unions are still talking about the future of AI through the lens of tools, cost savings, and FX jobs, like with discussions around AI and product development acceleration, we’re already moving past that moment. It’s not the future, it’s already the past. AI isn’t just a tool as we conventionally think of tools.

Edward makes the point this way (in response to Matt saying their new product Showrunner is a “replacement”):

A competitor is different to a replacement, but I think they they say, don’t worry, it’s just a pencil. Don’t worry, it’s just a paintbrush. I don’t know any pencils that start writing by themselves. So, I think people are highly intelligent, they can see through this completely, and I think honesty is better. And the honest truth is that this is creative by itself today, and that that is artistically very interesting, something Andy Warhol would have found completely fascinating, and it’s a new artistic medium, and it’s the first artistic medium that is aware and intentional. When people hear it’s just a tool, it’s just a pencil, they see through it. And it actually is more frightening because you think, what are you hiding from me? Like, if you’re really saying that, you know it’s not true, so you must be hiding something. So it’s better to be honest, I think.

This reality is absolutely going to blindside people. It’s the core of Matt Inman’s complaint about AI art (also rebutted rather nicely by John Gruber over at Daring Fireball.) It’s the reason to go beyond trivial interfaces and really think about AI as partners.

It’s the money, stupid

With Matt’s background as an entertainment lawyer, he is unusually crisp about intellectual property and guild issues. Edward and Fable are trying to frame their products as brand extensions and playgrounds. Syndication on steroids.

In the past, you’d have so much that you could distribute it 24/7. The new version of syndication is that once you have enough episodes, you can generate—unlimited is such a provocative term—but you can generate many more episodes and people can play in your show, and then it’s evergreen, it’s generating revenue for you on an ongoing basis.

Anyone want to bet against this being incredibly profitable for fans? So much smarter — and bigger in terms of TAM — than just being tools. Edward again:

And I think it, you know, the the path that a lot of these AI companies have gone down is trying to disrupt the VFX industry and actually they’ve raised more money, more money has been invested in disrupting the VFX industry than the size of the VFX industry, which I think says a reckoning is coming. That was not a good idea. It is not the right use of this technology. The right use of this technology is to embrace that it is not just a pencil, it can write itself, and it is creative. It’s not just a VFX tool and like a pipeline.

I’d quibble that the problem here is how you define a tool — and anything that can actually do real storytelling and show creation will, as a byproduct, completely disrupt VFX tooling — but sure.

This is going to apply to product development and coding agents in the same way. Like Hugh Herr notes with prosthetics, once technology matches human performance there is literally nothing stopping superhuman performance. Coding agents are chasing a similar inflection point — once they are producing durable code whole product development sectors will inevitably shift to agentic development and agentic co-creation.

It’s also going to transform storytelling in general.

Storytelling and bespoke metaverses

This week, Neal Stephenson announced a new Metaverse project, Artefact. Fun stuff and I’m glad he’s playing in the space, though it’s focus on crypto and blockchain feels like weirdly ancient technology to me. More than that, while there will likely always be players looking for MMO experience (hi, Raph and Stars Reach!) if I was thinking about metaverses I’d be starting at the intersection of Fable and companies like Character.AI, not legacy game engines or crypto.

Generative AI gives us some development superpowers and I expect the next wave of surprising experiences will take advantage of them.

• Medium mutability: To capture quotes from The Town, I used my vibe-coded transcriber because LLMs are incredibly capable of moving between text and audio. They’re super-human for translating many languages, too. As Showrunner is demonstrating, text to animation is going to be solved soon. Same with video as Sora 2, Veo 3, etc are demonstrating. Game experiences are actually easier but for Venn Diagram of “AI researcher” and “game developer” is apparently rarer than Demis and DeepMind would have you think. We are somewhere between “real soon now” and “working already” to allow an idea or experience to move between mediums with relative ease and high fidelity.

• Credibility, storytelling, and partnership: Without going deep into discussions of consciousness, we already have AIs that can credibly imitate a character in conversation and engagement with a user. Some of these engagements are ending tragically — and particularly when the engagement is with minors there is genuine complexity here — but people love being at the center of story. Games nearly always make you the hero, but they struggle to make the interactions in the game truly reflect you and your actions. GenAI and LLMs have the potential to make that possible. Moreover, think about all of the brittle finite state machine systems that could be tossed out in favor of an LLM that is focused on how you and your friends experience the game?

• Merging tool, platform, and experience: The fastest development teams are going to be the ones who embrace AIs in their development methodology, which means moving more and more development into the world of managing agents, prompts, and skills. Guess what? Not everyone using those agents, prompts, and skills will be programmers and once you cross that bridge, why aren’t users creating using the same tooling you are? Like Second Life 20+ years ago, Roblox and Fortnite are capturing the creativity and energy of their users — how much more exciting would AI tooling make this?

This, to me, is what the metaverses of the future are. Riff off an idea, an existing character, scene, topic, or idea. Create the right experience for what you want — whether it’s just a character to talk to or a full game to dive into with friends — and explore, share, and expand it. IP and ownership will matter a ton — who knows, maybe a public ledger would actually help here¹ — because it is critical to pay creators, especially if you’re trying to get major brands on board.

But being able to turn anything into a shared experience — persistent or otherwise — is the world that is coming very quickly.