Tim Shyu 中文EN日本語

Sam Altman's OpenAI Fantastic Journey

AI has progressed to the point where it's honestly hard to know what to say anymore. The most stunning development was OpenAI's event in early November, followed almost immediately by the firing of OpenAI CEO Sam Altman (as of this writing, he'd already been invited back onto the board, though nothing is settled yet) — and frankly, the two events are equally astonishing.

Let's start with just how stunning OpenAI's products have become. There's an old joke that's been making the rounds on Twitter: an investor asks a founder, "What's the difference between your AI model and OpenAI's?" The founder replies, "The difference is that ours is worse than OpenAI's." It's a lot like an old story about a fortune teller who tells a customer, "Before you turn 30, your life will be chaotic and unstable." The customer asks, "And after 30?" The fortune teller says, "You'll just get used to it." That's probably how a lot of AI researchers feel these days — they'd love nothing more than to beat OpenAI.

OpenAI's Apps Are Growing at Light Speed — Developers Are Feeling the Squeeze

It sounds pretty hopeless, but over the past six months plenty of people have actually been building models smaller than GPT that come close in some respects, or claim to be better or cheaper in others (Claude, for instance). These efforts have made real progress, and a lot of people figured there was a real commercial opportunity here — that we might be about to relive the Web 2.0 era.

But then OpenAI's latest event unveiled GPT-4 Turbo and the GPTs tool, and overnight, that joke started to feel true again. OpenAI's applications have already shown a future where ordinary people can build complex tools of their own. You don't need to know how to code — you can use natural language to build all kinds of handy tools, something that used to be unthinkable. It's still a child today, sure, but you can already tell it's going to grow up to be six foot seven and dunking on everyone.

My takeaway: they should rename Developer Day to "You Don't Even Need to Develop Anymore Day." OpenAI is, without question, the leader of the pack. There are plenty of companies doing cutting-edge technical work in this space, but very few that can make a random passerby instantly grasp where AI is heading. The entire keynote was delivered in plain, non-technical language, describing what the new GPT tools could do: they used GPT to plan a trip to Paris, to build a daily reminder tool, and so on — just like Jobs, hiding the technology beneath a surface anyone can understand, and lifting the curtain just enough for you to sense the sophistication underneath.

This is a kind of democratization of AI, a process of making it accessible to everyone. The definition of "developer" has now expanded enormously — plenty of people who barely know how to code can now put together polished applications. So even though it's called a "Developer Conference," the actual message is basically: everyone, please go lie down and relax — there's absolutely no expectation that developers sprint to build something in some narrow window of time.

It used to be that a month or two after a launch, you'd see a wave of new applications built on top of it — and "a wave" usually meant a few hundred to a couple thousand. Now, just ten days after the announcement, there are already more than 14,000 GPT-based apps you can find on review sites. The funniest part is that there are now so many "review sites" that there are review sites reviewing the review sites! Developers are going to face enormous pressure to keep updating and to pick a side, and as this cycle of updates keeps accelerating, most ordinary developers are going to get squeezed out at an even faster pace.

But then an even more shocking situation emerged: the CEO who had just led the company to the breakthroughs described above, Sam Altman, got fired! Then, as of November 19 Taipei time, it seems investor pressure got him invited back. It was essentially a sudden palace coup, followed by an equally sudden counter-coup that ousted the original board. Word then went around that Emmett Shear, founder of game-streaming platform Twitch, had been installed as the new interim CEO, before Microsoft CEO Satya Nadella finally announced where Altman had landed: he'd joined Microsoft, the very company that had invested in OpenAI. The coup, it seems, has come to an end.

This whole sequence of events leaves you breathless — and calling Altman the Steve Jobs of our era wouldn't be an exaggeration at all. Just like Jobs back in the day, Altman was fired by the board of the very company he founded, only to zip right back at light speed into the same orbit as Microsoft. It's a genuinely fascinating situation. Right now AI is having something like an "iPhone moment" — the pace at which applications are multiplying barely needs explaining. When the iPhone first launched, all of 2008 produced just over twenty thousand apps total; today, that number can easily be hit in a single month. So you could say Altman is the Steve Jobs who got fired right around the launch of the iPhone 3G. That a board can just casually stage a coup like this makes you feel like this company's governance isn't so different from a college club's.

Former CEO Altman Joins Microsoft — Board Coup Comes to a Close, For Now

As for why Altman was fired, what's known so far is that it was a surprise move by OpenAI's internal safety-hardliner faction, who believe AI safety isn't yet good enough and that the issue needed to be addressed immediately. Altman appears to represent more of a "ship it and see" development-first camp. Both sides have their logic, but just how serious a safety concern would have to be to justify a board coup on this company-destroying scale — that still needs more information to answer.

For now, the fallout is a huge win for Microsoft — a company with genuine corporate governance has effectively stepped in to manage Altman and everyone else. Going forward, OpenAI-related matters may well become more stable. But what exactly was the safety issue serious enough to justify casually firing the founder? That still needs to be clarified, and it's presumably not something simple. AI regulation may be a bigger issue than we've assumed, and more people are likely to start paying attention to it going forward.

All of this also makes you marvel at just how fast the AI era moves. Take the Steve Jobs biography or the American TV show Silicon Valley (矽谷群瞎傳), play it at 30x speed, and that's basically the show we just watched play out. Three CEOs in three days — not even a TV writers' room would dare write that!

The Tim Shyu Letter

First-hand notes on AI agents, marketing tech, and the content industry — straight to your inbox.

Subscriptions open when the site goes live. Hold tight.


← All articles