OpenAI's chief technology officer, chief research officer and a number of key researchers have abruptly quit. Those departures were just the latest in an ongoing exodus of key technical talent. Over the past few years, OpenAI has lost researchers who played crucial roles in developing the algorithms, techniques, and infrastructure that made it the world leader in AI. Of 31 people listed as authors of OpenAI’s GPT large language model, fewer than half remain. New applicants are drawn to work with a particular researcher or team, and their calculations could change as more big-name researchers leave for rival AI companies or their own startups. A question is whether OpenAI has a deep enough bench to continue to be first in AI, essential for its skyrocketing valuation.
When investors poured $6.6 billion into OpenAI last week, they seemed largely unbothered by the latest drama, which recently saw the company’s chief technology officer, Mira Murati, along with chief research officer Bob McCrew and Barret Zoph, a vice president of research, abruptly quit.
And yet those three departures were just the latest in an ongoing exodus of key technical talent. Over the past few years, OpenAI has lost several researchers who played crucial roles in developing the algorithms, techniques, and infrastructure that helped make it the world leader in AI as well as a household name. Several other ex-OpenAI employees who spoke to WIRED said that an ongoing shift to a more commercial focus continues to be a source of friction.
“People who like to do research are being forced to do product,” says one former employee who works at a rival AI company but has friends at OpenAI. This person says some of their contacts at the firm have reached out in recent weeks to inquire about jobs. OpenAI itself has also seemingly shifted in its hiring priorities, according to data compiled for WIRED by Lightcast, a company that tracks job postings to analyze labor trends. In 2021, 23 percent of its job postings were for general research roles. In 2024 general research accounted for just 4.4 percent of job postings.The brain drain could have lasting implications for OpenAI’s direction and future success. Experts and former employees say the company still has a deep bench of talent, but competition is intensifying, making it more challenging to maintain an edge.
The latest big-name departure, revealed on Thursday, is that of Tim Brooks, head of OpenAI’s Sora AI video generation project. Brooks posted on X that he would join one of OpenAI’s main rivals, Google DeepMind.
“It could start to change things,” says a former OpenAI staff member, who now works in academia, of the losses. They asked to remain anonymous out of concern for harming collaborative relationships with the AI industry.
For now, this person says, many students still put OpenAI at the top of their list of potential employers. It is seen as several months ahead of the competition, and prospective employees are often willing to put up with the apparent drama and infighting to be part of that. But applicants are also often drawn to working with a particular researcher or team, and their calculations could change as more big-name researchers leave for rival AI companies or their own startups.
A look at some of OpenAI’s most important research shows how much talent has departed. Of 31 people listed as authors of an early version of OpenAI’s GPT large language model, fewer than half remain at OpenAI, according to employment details sourced from LinkedIn or other public social media profiles. Several members of the team responsible for developing GPT left OpenAI in 2021 to form Anthropic, now a major rival. Roughly a third of those listed in the acknowledgements for a technical blog post describing ChatGPT have since left.A key question is whether OpenAI has a deep enough bench of AI researchers that it can continue to be first to key advances in AI, essential for justifying its skyrocketing valuation.
OpenAI aims to increase its revenues to $100 billion per year by 2029, according to a report in The New York Times that cites a document shared with prospective investors. The company expects ChatGPT to bring in $2.7 billion in revenue this year, up from $700 million in 2023, with a further $1 billion coming from other sources. Competition is also intensifying, with Google and others throwing huge resources behind competing AI offerings. The social media giant Meta offers free AI models and tools that are often as good as those provided by for-profit players.
Some of the biggest shoes for OpenAI to fill are those of Ilya Sutskever, a cofounder and the technical visionary for much of the company’s early work. Sutskever departed in May to found a new AI company of his own called Safe Superintelligence.
Sutskever did pioneering work on deep learning and was an early believer that scaling up the amount of compute used to build models would yield increasingly capable AI—a prediction that proved remarkably prescient. He also crucially realized that a neural network architecture developed by Google researchers known as the transformer could prove key to improving AI’s language skills.
Sutskever was also a member of the board that briefly ousted CEO Sam Altman in November 2023, although he apologized and called for Altman to be reinstated a few days later. The episode seemingly stirred up internal conflict over Altman’s control of the company, its commercial focus, and appetite for risk. Murati was often caught in the center of internal conflict, according to a report in The Wall Street Journal, including tensions over the hastiness with which the company’s voice interface was released.
Sutskevar’s departure was swiftly followed by that of Jan Leike, who co-led a team dedicated to managing the long-term risks of AI. In August John Schulman, another OpenAI cofounder and a highly respected research scientist, also announced that he was leaving. Schulman worked on, among other things, techniques for fine-tuning large language models with human feedback using what’s known as reinforcement learning, an approach that proved key to building a capable and garrulous chatbot like ChatGPT. Another big name to leave this year was Andrej Kaparthy, a cofounder who rejoined the company in 2022 after a stint as director of AI at Tesla. Kaparthy left in February of 2024 to found Eureka Labs, an education AI startup.
Another early OpenAI employee, who spoke anonymously because they were not authorized to discuss the matter by their current employer, says these departures are significant because of how important the research work produced by the individuals has been to OpenAI. However, this person notes that OpenAI still has a number of extremely talented young researchers—and can also offer huge sums of money to potential recruits thanks to its soaring valuation.OpenAI’s latest round of investment comes as the company works to turn itself into a for-profit company. The original founders included Elon Musk, who provided $45 million in early funding. After Musk parted ways with the company, Altman turned it into a capped for-profit enterprise overseen by a board of directors that does not answer to investors. Last November’s attempted coup proved this structure to be in direct conflict with a fast-growing, heavily backed technology business.
One former OpenAI staffer, who similarly asked to remain anonymous, confirmed that the product focus continued to cause tension, and said they had been told by current staff to expect more turmoil at the company.
Other ex-staffers say the picture inside the company is complex. “The mood is bleak for some and not for others,” one person said.
When asked about the situation, OpenAI referred WIRED to a message that Altman sent to staff after the departures of Murati, McGrew, and Zoph, which he also shared on X. In the note Altman names several researchers who will step into more senior roles, including Mark Chen, who will become a senior vice president of research, and Jakub Pachocki, who becomes the company’s chief scientist.
“Leadership changes are a natural part of companies, especially companies that grow so quickly and are so demanding,” Altman wrote. “I obviously won’t pretend it’s natural for this one to be so abrupt, but we are not a normal company.”
“If you think that a few individuals have some special sauce, and that has just walked out the door, then that's a problem,” says Simon Johnson, a professor at MIT’s Sloan School of Management.
However, Johnson says that OpenAI’s loss may well turn out to be a gain for the rest of the tech industry.
“I think we need a lot more competition in AI; we need a lot more business models; and we need a lot more smart people coming up with constructive things,” he adds. “So I'm very comfortable with the people leaving.”
When OpenAI’s board ousted CEO Sam Altman last November, staff threatened to quit in protest, and many shared the slogan “OpenAI is nothing without its people” on X. The company seems determined to test how replaceable some of its key people really are.
0 comments:
Post a Comment