How AI agents "radicalized" a top Meta exec into quitting her job
Platformer Casey Newton ● Covered by 3 sources
A top Meta AI exec quit after seeing agents do work that used to take a whole team. She says the scary part isn’t coding — it’s how fast entry-level jobs can shrink.
Based on reporting by Platformer, Casey Newton — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
Clara Shih didn’t leave Meta because she lost faith in AI. She left after watching it work too well.
While leading Meta’s business AI group last fall, she saw AI agents compress a product process that once took user researchers, designers, product managers and three kinds of engineers into one or two people and a prototype. The same pattern showed up in marketing, distribution and privacy review. Tasks that used to take 10 steps and 10 days could suddenly be pushed through in minutes, with humans still checking the final output.
That was the moment, she told Platformer’s Casey Newton, when the idea stopped being abstract. She had already started pulling down entry-level job postings at Meta because she no longer felt they were needed. This spring, she left the company, stayed on as a senior advisor, and started the New Work Foundation and Dear CC, a free set of tools aimed at entry-level workers.
Shih’s view is not the clean, comforting one that AI boosters like to sell. She says the old story — automate the repetitive stuff and workers will naturally move up to better work — has “primarily not been true.” She also thinks the impact won’t be spread evenly. Some jobs will be directly substituted. Some will get better if the routine parts vanish. And some will still exist, but with the barrier to entry lowered so much that the job turns into something worse.
She’s especially focused on roles where people prepare something for someone else to review: briefs, slide decks, order forms, and similar work. Her rough count is that about one in five corporate roles falls into that bucket. Those jobs, she says, are the ones most exposed when the person at the end of the chain can use AI to skip all the handoffs and get to the answer faster.
The part that should make companies uncomfortable is that Shih isn’t speaking from the outside. She helped build Salesforce’s Agentforce and Meta’s Business AI. She saw the machine from the inside, then decided to build a nonprofit around the people it may squeeze first. That is a much less cheerful takeaway than “AI will create new jobs.” It’s also the one that sounds like it came from somebody who actually watched the gears turn.
My take — AI-written commentary, not fact-checked reporting
The real tell here is how quickly “AI for productivity” turns into “we probably don’t need those junior jobs anymore.” That’s not disruption; that’s a hiring strategy with better branding. The tech industry has spent years pretending every efficiency gain is automatically a public good, which is cute, if you enjoy fiction.
Read more about this at: Platformer