OpenAI Returns to ‘Normal;’ the AI Marketplace Is Changed Forever

Credit: Adobe Firefly

I don’t mean to dwell on the OpenAI fiasco. But I did, like everyone else covering AI even tangentially, chime in on it, so it feels appropriate to draft an epilogue to the whole affair, at least as I see it. While I’m just as riveted as everyone else by the soap opera, I sincerely hope this is the last time I talk about the failed OpenAI coup, at least directly.

As everyone knows, Sam Altman is returning to OpenAI along with Greg Brockman, and a new board of directors is under construction. Microsoft and Emmett Shear seem satisfied, if not entirely happy, with how everything turned out. We never got any definitive answers on exactly what motivated the board to fire Altman, but Reuters reported a breakthrough step toward artificial general intelligence (AGI) codenamed Q* (pronounced Q-star) may have been the catalyst.

Does that mean this was all about a split between accelerationists and altruists, where the board thought it was doing its job by kneecapping the company to slow down the progress toward AI being some kind of threat to humanity? Does Q* meaningfully move OpenAI closer to AGI, or is it all just hype to intrigue customers and scare competitors? Is Microsoft the big winner, or did they come out worse than before?

And most importantly: Who cares?

But really, who does care? Silicon Valley observers, certainly. Investors, definitely. People who have a stake in the idea of AGI, whether they’re in the safety camp or the move-fast-and-break-things camp, completely. OpenAI’s competitors, wholeheartedly.

But for everyone else who doesn’t work in the Bay Area and whose existence doesn’t hinge on whether we’ve achieved AGI, the new twists in the story — while certainly intriguing and potentially market-moving in the long term — don’t matter a lot. As monumental AGI might be to the human race, it actually has very little to do with the day-to-day of anyone building, deploying or working with AI tools today.

All the revelations about the board and whispers about Q* haven’t changed the takeaways for people working with generative AI tools. However, how much those takeaways alter your approach depends on how much you’ve invested in OpenAI’s tech. What to consider:

  1. AI is a very young technology.

  2. While AI is generally a very portable technology — tools can leverage any number of foundational models — fine-tuning isn’t. The model you choose matters, and you should think twice about choosing one you don’t control.

  3. OpenAI has a first-mover advantage and large name recognition, but it is a very unusual company, which creates unusual risks (although it looks like those are being actively reduced).

  4. The most responsible thing an AI user or developer can do in the wake of OpenAI’s crisis is to diversify their AI toolkit.

Ultimately I believe what’s happened will be healthy to both the AI ecosystem and OpenAI itself. After this period of instability, OpenAI will need to work hard to win back the trust of their customers and the developers building on its platform. At the same time, competitors — especially Anthropic, which arguably has the second-most-capable general-purpose AI chatbot — will rise. 

Subscribe now

An App Analogy

How that will all shake out is anyone’s guess, but it may be useful to think about an analogous situation in a parallel universe: Consider if, on the eve of the launch of the iPhone App Store in 2008, there was a coup at Apple to oust Steve Jobs as CEO. It would never have happened, certainly; Jobs was too smart about his hires over the previous decade, but it’s an interesting thought exercise.

Even if Jobs staved off the coup and quickly won back his CEO title in a manner similar to Sam Altman, you have to think the incident would have damaged Apple. The release of the Android Market and the first Android phone, the HTC Dream, both of which happened a few months later, may have had a lot more customer interest and developer energy behind them after Apple had shown itself to be so chaotic and vulnerable.

The analogy isn’t perfect. Apple had a long, storied past and a cult-like following thanks to so many successful products; OpenAI is really just beginning its journey as a consumer brand. And Apple’s affluent customer base was attractive to developers regardless of what was going on with Android.

But that’s all the more reason to believe OpenAI’s real problems are even worse than this hypothetical Apple’s fictional ones. Whereas phone ecosystems are hard to just pick up and put down, AI developers have their pick of foundational models. While they wouldn’t in their right mind shun ChatGPT’s large customer base, after the past week they’d be just as crazy to attach their destinies to a single vendor — especially one that came within a stone’s throw of imploding.

The Media Copilot is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber.

Ready to start using AI like a pro?


Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.