OpenAI enters 2024 beset with unanswered questions
ChatGPT maker OpenAI weathered an epic leadership crisis nearly two months ago but enters 2024 beset with uncertainties.
Why it matters: The company that kicked off and still leads today's AI boom remains an enigma — it could be the next Google, or the next Netscape.
Zoom in: OpenAI's "openness" has some big limits that make it hard to forecast its near future.
- Its biggest secret right now is what data it has used to train the AI services it offers, particularly GPT-4, the large language model that powers ChatGPT.
- If the New York Times' copyright lawsuit against OpenAI goes to trial, the company could have to divulge more details about its bots' data diet.
- AI firms hold this information close for competitive reasons, but critics suggest that OpenAI's nonprofit mission to ensure that AI "benefits all of humanity" means it should be held to a higher standard.
- After reconstituting its board with three members, one holdover and two newcomers, OpenAI has promised to expand the governing body — but has not yet named any additional directors.
- Microsoft executive Dee Templeton has begun attending OpenAI board meetings as a nonvoting observer representing Microsoft's interests, per Bloomberg.
There's no news from OpenAI about the independent investigation into the November board crisis that the firm promised after reinstating Altman.
- One root of the conflict, as Axios reported, was that some board members felt they could no longer trust Altman and had concerns about his projects and investments outside OpenAI.
- But the board never backed up those concerns with evidence or details, leaving Altman with the overwhelming support of OpenAI employees.
The November crisis also triggered reports that board members were spooked by major new breakthroughs in AI development that might bring the field closer to the kind of existential crisis OpenAI had been founded to prevent.
- Since the conflict's resolution, however, those reports haven't been corroborated.
The big picture: The toughest questions hanging over OpenAI aren't existential, but financial.
- A former employee once described the company to Platformer's Casey Newton as "a money incinerator."
- The firm's giant AI models, like GPT-4, cost a fortune to develop, train and run.
- That's why Altman and OpenAI have had to raise tens of billions of dollars from investors led by Microsoft — and why the company, nominally organized as a non-profit, operates more like a traditional Silicon Valley startup.
- OpenAI is now seeking more investors who are willing to value it at $100 billion or more, per Bloomberg. If the AI bubble cools in 2024, the firm could face a reckoning.
Tech history tells us that the company that kicks off a boom isn't always the one that survives to reap its rewards.
- Netscape's 1995 IPO ushered in the internet era, but by 1998 the browser pioneer was on the ropes. In 1999 it got bought out by AOL, which also crashed and burned — leaving Google and Facebook to harvest the web's profits.
- Similarly, before there was an iPhone, there was Blackberry. But despite the loyalty many users felt for the Canadian innovator's products, they became footnotes once Apple's touch-screen wonder took off.
Yes, but: Right now OpenAI has a broad customer base that's willing to pay $20 a month to use ChatGPT. And it has everyone's attention.
The bottom line: The startup at the crest of each new wave in tech can end up on top of the world — or buried in an undertow.