Meta Used Its Employees as AI Training Data, Then Laid Them Off

A leaked audio recording reveals Zuckerberg's real explanation for why Meta was tracking its workers.

You've heard of "learn by doing." Meta invented "learn by watching employees do it, then replace them."

A leaked audio recording from a Meta all-hands meeting, obtained by More Perfect Union, captures Mark Zuckerberg explaining that Meta has been monitoring employee activity across Gmail, GChat, internal tool Metamate, and VSCode (the coding software most engineers use) to train its AI models. His reasoning: the AI "learns from watching really smart people do things," and elite engineers make better training subjects than outside contractors.

April 21: Meta announced that it was installing keystroke and mouse-tracking software on employee computers. Meta's public response was mild: the models just needed to learn how people click dropdown menus. Routine stuff.
April 30: At an internal all-hands, Zuckerberg gave a rather different explanation on tape. The AI learns from watching "really smart people." He also acknowledged the leak risk, telling staff it was "not strategically in your interest" to share details openly.
May 19 (Monday): Meta announced it was reassigning 7,000 workers to new AI-focused teams. Framed internally as a productivity upgrade.
May 20 (Wednesday): Roughly 8,000 employees received layoff notices, starting at 4am Singapore time. The leaked audio dropped the same day. Fliers opposing the tracking program appeared on office walls.

The practice of hiring cheap labor and requiring employees to "train their replacements" has been going on for a long time. The quality of that "training" was always questionable. It was extorted on pain of receiving a severance with your layoff.

During my time at PacBell, we created an improved data entry system that employed our own employees instead of farming out the work to data entry personnel offshore. There was a requirement in the development of the software to demonstrate improved productivity over the farmed out alternative. After a month, the productivity information was extracted. It was detailed to each employee. A surprising result was that the employee perceived to be the most productive was in fact entering the same data over and over to look "busy". The employee perceived to be the slowest was actually doing quality, in-depth research and data entry and was the most productive person. The person trying to look busy received a stern warning. There was an unhappy feeling of having been "spied upon" by the company when the productivity details were extracted. In the end, the software was far better than the farmed out labor and the internal employees work quality was more highly valued.

What Zuckerburg did with AI watching his employees to "learn how to do the job", then tossing them on the street is a pretty low life action. My hope is that the AI only saw a subset of what the real employees can do and Zuckerburg finds himself with a half-assed AI incapable of doing the whole job and employees that won't have anything to do with him again.

I was sent to Sacramento to learn how the employees read the SORD SOIR (Service Order Interface Records) and entered the data into the system that I was standing up. Everything necessary was in the SOIR record. It was just a matter of parsing it and generating an input record into my own support system. It was done in 3 days. Ready for production. Lots of manual work removed. Certainly not AI, but working in the spirit of automating work that was suited for automation.

Private property is under assault. Any written work is fair game to these Ai developers. Behind a pay wall or not. They call it “scraping the internet.” They want more and more Data. Private ownership of your product, copyright infringement be damned. Information acquisition is everything. If someone eventually sues, well, they already have billions. They will settle out of court, or countersue the original author and tie them up in court.

In a 1973 interview, Vonnegut discussed his inspiration to write the book Player Piano:

I was working for General Electric at the time, right after World War II, and I saw a milling machine for cutting the rotors on jet engines, gas turbines. This was a very expensive thing for a machinist to do, to cut what is essentially one of those Brâncuși forms. So they had a computer-operated milling machine built to cut the blades, and I was fascinated by that. This was in 1949 and the guys who were working on it were foreseeing all sorts of machines being run by little boxes and punched cards. Player Piano was my response to the implications of having everything run by little boxes. The idea of doing that, you know, made sense, perfect sense. To have a little clicking box make all the decisions wasn't a vicious thing to do. But it was too bad for the human beings who got their dignity from their jobs.

Regards,

Disclaimer: Opinions posted on Free Republic are those of the individual posters and do not necessarily represent the opinion of Free Republic or its management. All materials posted herein are protected by copyright law and the exemption for fair use of copyrighted works.