Thanks for all the hard work everyone, but we’re gonna let the AI agents take it from here.
Driving the news: Meta will begin tracking its U.S. employees’ mouse movements, clicks, keystrokes, and take random screengrabs of their computers as part of an initiative to build out AI agents that can perform workers’ jobs autonomously, per Reuters.
Speaking about the new employee tracking tool, Meta’s chief technology officer wrote in a memo, “The vision we are building towards is one where our agents primarily do the work and our role is to direct, review and help them improve."
CEO Mark Zuckerberg has already built an AI agent to be his co-CEO and is reportedly working on a 3D animated avatar of himself for employees to interact with (not creepy at all).
Catch-up: Human data is essential to building top AI models, so much so that Meta paid US$14.3 billion for a 49% stake in Scale AI, a startup that specializes in turning human-generated data into training material for AI models. Scale AI’s founder, Alexandr Wang, is now leading Meta’s AI superintelligence team.
Why it matters: Meta might be early to the AI agent party, but other companies won’t be far behind. The real question is, once these white-collar workers have trained the AI models on how to do most of their job, what’s stopping them from being laid off?
If workers truly become de facto babysitters for an army of AI agents (as Meta envisions) it seems likely companies would significantly cut their workforces.
Yes, but: There’s still a lot of risk in outsourcing too much responsibility to AI tools. To name a few recent blunders: Air Canada’s autonomous AI bot offered a customer a refund policy that didn’t exist, Meta’s in-house agent went rogue and compromised large sets of sensitive customer data, and just this week, the Wall Street firm Sullivan & Cromwell admitted that its filing in a high-profile court case was riddled with errors thanks to AI hallucinations.—LA
