Who’s actually getting rich off of AI?

Today, let’s talk about a subject that crosses my mind with every generative-AI startup pitch that lands in my inbox: who’s going to make the real money off artificial intelligence?

Last week the productivity startup Notion announced that Notion AI, a suite of tools based on OpenAI’s ChatGPT, had entered general availability. For $10 per user per month, Notion can now summarize meeting notes, generate lists of pros and cons, and draft emails.

Notion AI is among the first in a wave of companies that are racing to capitalize on growing interest in generative AI. This week Snapchat made available a ChatGPT-based chatbot called My AI for subscribers to its $4-a-month Snapchat Plus offering. The educational app Quizlet announced a ChatGPT-based tutor called Q-Chat. And Instacart said it is developing a tool that that will let customers ask about food and get ‘shoppable’ answers informed by product data from the company’s retail partners.”

What I’m interested in, as more and more companies adopt features like this, is where the ultimate value lies. Will an ever-growing number of companies find ways to integrate AI into products that are valuable enough to charge for — or will the bulk of the profits go to the small number of companies building and refining the underlying models on which those tools are based?

The answer will go a long way in determining whether generative AI represents a true platform shift on the order of the move from desktop computers to mobile phones, or a more limited set of innovations whose benefits accrue to a handful of big winners.

The subject is clearly on developer’s minds. This week, in response to concerns, OpenAI said it would no longer use developers’ data to improve its models without their permission. Instead, it would ask developers to opt in.

“One of our biggest focuses has been figuring out, how do we become super friendly to developers?” Greg Brockman, OpenAI’s president and chairman, told TechCrunch. “Our mission is to really build a platform that others are able to build businesses on top of.”

Maybe it’s that simple — developers don’t want to help OpenAI refine its models for free, and OpenAI has decided to respect their wishes. This explanation feels more consistent with a world where AI really does represent a platform shift.

Or maybe OpenAI believes that its models can continue to improve rapidly with or without all of those developers opting in. This explanation feels more consistent with a world in which OpenAI and a handful of others reap most of the rewards of AI.

So what kinds of AI features are people actually selling?

For the moment, AI products on the market are essentially just white-labeled versions of ChatGPT. As of this week, OpenAI is making it available to other companies through an API. For $0.002 for about 750 words of output, any company can resell ChatGPT in their own app.

For the moment, then, there’s not really much consumer choice when it comes to generative AI. To the extent that there are multiple options, it’s in interfaces. Do you want to draft an email using AI? It might be more convenient in Notion, where you already have some meeting notes. Do you want to get some recipe ideas or ask a quick trivia question? If you’re away from your laptop, it might be fastest to ask My AI on Snapchat.

For the moment, there are still billions of people who have never used ChatGPT. Introducing a re-skinned version of that service to a popular consumer app like Snapchat, which has 750 million monthly users, could help it find a whole new audience. Paying Notion or Snapchat for the feature also guarantees access to a service that has often gone offline amid heavy usage of its own web app.

At the same time, services like this are largely just making a bet against copy-paste. You can already get essentially everything here for free inside ChatGPT; apps like Notion and Snapchat are selling what feel like a fairly minor convenience for a significant premium.

The promise is that these tools will become more personalized over time, as individual apps refine the base models that they rent from OpenAI with data that we supply them. Every link that has ever been in Platformer is stored in a Notion database; what if I could simply ask research questions of the links I have stored there?

Those sorts of features are coming, Notion CEO Ivan Zhao told me in a recent interview. The initial set of writing and editing tools represent a “baby step,” he said. But much more profound changes are coming.

“I’ve never been this excited about something,” said Zhao, who is not prone to hyperbole. “It feels like electricity: the large language model is the electricity, and this is the first light-bulb use case. But there are many other appliances.”

Zhao told me that Notion would add lasting value to the language model by creating interfaces that are easier to use than rivals’. This is a more important project than it may first appear: one reason voice interfaces like Siri and Alexa have plateaued is that users struggle to remember all the different things they can be used for. (Another is that their language models aren’t nearly as sophisticated as ChatGPT’s.)

“This is as much an interface revolution as a technology revolution,” Zhao told me. “And we’re good at interfaces.”

There should also be real value in more personalized AI models. I’ve spent the past year or so writing a daily journal in an app called Mem, which offers its own set of ChatGPT-based features to premium subscribers. Eventually, I imagine I’ll be able to ask my journal all sorts of questions in natural language. What was I worried about last summer? When’s the last time I saw my friend Brian? A journal that gets good at that sort of thing could command a premium price, I think.

Still, there’s probably a limit to how many add-on AI subscriptions most people will want to pay for. Over time, the cost for these features seems likely to come down, and many of the services that cost $10 a month today may someday be offered for free.

But that, too, poses a risk for startups betting on features like these to drive growth. The cheaper the services get, the more likely they are to be offered for free by big platforms. What happens to Notion when its full suite of premium AI tools is offered for no cost within Google Docs?

The best-case scenario for these companies, I think, is that generative AI comes to resemble the cloud-computing market. The basic infrastructure will be built by a handful of companies, but the capabilities they make available cheaply inspires an entire new generation of startups.

At the same time, the capabilities of ChatGPT and other large language models are only going to expand — and I suspect they may gobble up a lot of today’s AI startups along the way. As they implement AI features into their apps this year, tech companies would do well to ask whether they might be building on top of a sinkhole.