In the computer sector, where employment is generally dire, there were stories last month of some incredible job offers. Packages totaling hundreds of millions of dollars were being used to entice AI researchers. A.I. free agent recruitment has turned into a social media show, similar to the time leading up to a trade deadline in sports, according to the New York Times, which compared the hiring tactics of the business to those of the NBA.
The following revelations have served to clarify the causes and validate what the original accounts simply hinted at: Star AI engineers are in high demand, but just one company—Meta—is making genuinely remarkable deals. Naturally, other businesses are vying for the attention of their AI talent, but none of them are matching Mark Zuckerberg’s nine-figure offers, and some are even declining to make any counteroffers. For what reason?
Meta is an odd reputational outlier among the constellation of large tech companies (Google, Microsoft, and Meta), established startups (OpenAI and Anthropic), productless new ventures (Safe Superintelligence, Thinking Machines Lab), application companies (Cursor, Replit), and, uh, “other” firms (xAI) vying for the limited number of individuals with knowledge and experience in cutting-edge AI development: A business that appears to need to outbid its rivals by multiples and has lofty goals and practically limitless resources
On the surface, the narrative isn’t particularly intriguing. Even though there was significant talent at the time, Meta’s shift to AI came later than other large tech businesses, partly because the corporation had just made the switch to the “metaverse.” As a result, Meta is clearly farther from the action than the others. Meta is obviously in pursuit mode, in contrast to OpenAI, which was the first to transform current LLMs into a product, or Google, where a large portion of the research that sparked the recent AI boom was conducted. Its models aren’t competitive, above all.
Additionally, Zuckerberg’s recent, Altman-style attempts at AI prediction are even less credible from within the industry than from outside. To the extent that he has attempted to articulate a vision—replacing friends with AI, a nebulously defined “superintelligence” that will somehow primarily benefit individuals—it comes across as borrowed and incoherent, even by the low standards of CEO AI prognostication. Furthermore, the talent that AI companies are seeking tends to be fairly idealistic by certain standards, regardless of the reasons you may assign to its leaders. If they are concerned about the risk posed by AI, they are likely more likely to work for someone like Dario Amodei of Anthropic than Mark Zuckerberg in 2025.
If they are interested in pure research, they will likely wish to work with people who share their interests, such as Ilya Sutskever, the creator of Google, or Demis Hassabis of Google. They could be more interested in working at OpenAI if they are a little more accelerationist and spend a lot of time reading the kind of AI debate that is popular on X, or if they are just examining areas where like-minded talent is already concentrated. Even xAI has a compelling argument for a specific kind of “cracked” AI developer, despite some of its gory aspects: plenty of resources, few restrictions, close ties to Elon Musk, and a goal that aligns with it.
Perhaps most significantly, despite different beliefs and distinct reasons about what the AI arms race actually is, few individuals in the business would wager on Meta winning it—at least not at this time. (However, one may contend that the company’s entertainment-focused, slop-and-advertising-related strategy for monetizing AI thus far offers a practical, albeit disappointing route to obtaining value from LLMs.)
The fact that Mark Zuckerberg’s company’s outlier proposals are mostly explained by the fact that his future-of-AI presentation is weaker than the MechaHitler guy’s (recent reports about internal documents permitting “sensual” talks with youngsters surely can’t help). The reputational aspects of Meta that are less well-known outside of the IT sector, however, are nevertheless very important. Richard Nieva and Rashi Shrivastava of Forbes discuss more serious issues, including ongoing internal reorganizations (the fourth in six months is reportedly under way), hazy objectives, and grueling pressures to work on products that haven’t been well accepted thus far.
Employees at AI startups are frequently driven by real excitement, fear of their companies losing a high-stakes race, and, of course, a great deal of stock attached to the success of their work, even as severe “9-9-6” work patterns are becoming more prevalent. Since 2023, Meta has taken a more harder stance toward its workers overall, with leadership publicly warning the public and implementing layoffs that significantly deviate from the company’s financial results.
A (now-former) AI researcher at Meta issued a widely shared internal note warning that the company’s work culture was in danger due to a “metastatic cancer.” He explained his goal to highlight the “unhealthy” aspects of the AI group in a post last week that was intended to counteract “sensationalist” reporting on the letter, but it wasn’t any kinder:
• The everyday anxiety that people have about being reviewed and dismissed, which is a major source of morale-boosting remarks.
• The requirement for a culture and procedures that make “big projects” possible.
• A lack of belonging among employees is caused by the management culture’s failure to foster a sense of camaraderie.
• Unstable team assignments that prevent experience from accumulating and solidifying over time.
• The crew found it difficult to passionately come together around a shaky vision.
This is very damning information for a corporation that was long thought of by people in the field as a decent place to work, with high pay, nice benefits, and, if not a fantastic culture, at least a productive one. Among other things, it implies that management’s attempts to discipline its employees may have had unanticipated consequences and that the bill for Meta’s ongoing attempts at reinvention may be approaching due. This indicates a negative outlook for Meta’s AI endeavors and provides a somewhat different narrative than what the company’s flurry of massive offers first suggested: Top AI recruits are no longer expected to accept billion-dollar offers. As it happens, they are Meta’s sole weapon in the talent battle.
