Google terminates 200 AI contractors — 'ramp-down' blamed, but workers claim questions over pay and job insecurity are the real reason behind layoffs
Some believe they were let go because of complaints over working conditions and compensation.
Google has laid off over 200 contractors who worked on improving its AI product offerings, Gemini, and search AI overviews, according to Wired. Some were told it was part of a "ramp-down" on the project they were working on, but others believe it was due to complaints made over pay and working conditions. These laid-off contractors join hundreds of other AI-related contractors who have been fired from other major AI firms like xAI and Meta in recent months.
For the first half of 2025, AI growth was everywhere, and all the major companies were spending big to try to get ahead. Meta was offering individuals hundreds of millions to join its ranks, and entire companies were swallowed up in the race to be the first to the next big development in AI technology. But while announcements of enormous industry deals continue, there's also a lot of talk of contraction, particularly when it comes to lower-level positions like data annotation and AI response rating.
In the case of Google's latest firings, the contractors mostly worked at GlobalLogic, a software developer owned by Hitachi. Per the report, workers on the 'super rater' program were required to have either a PhD or a master's, and were tasked with moderating AI responses.
Google seemingly didn't need that expertise any longer, though. It has reportedly been working on developing an AI model for rating AI responses, though it's not clear if that's matured enough to take over what the human raters were doing. The Wired report even cites some workers worried "they are being set up to replace themselves."
Instead, many of them believe that it is their complaints over compensation that lead to them being laid off. Wired states that workers "attempted to unionize" earlier in the year to no avail. According to the report, "they allege that the company has retaliated against them." Two have filed complaints with the National Labor Relations Board. For its part, Google said in a statement that GlobalLogic is responsible for the working conditions of its employees.
AI developmental companies have been repeatedly accused of using underpaid human workers to classify AI data and rate responses to improve their services. The company Meta invested almost $15 billion to poach their CEO, ScaleAI, was accused of using low-paid workers in Southeast Asian countries to build out its data annotation services. After Meta bought a 49% stake in the company, over 200 of its staff were cut.
Just a few days ago, xAI also announced it was laying off around a third of its data annotation team, amounting to over 500 workers. However, unlike Google, company messaging suggests this is to pivot towards more specialist AI trainers.
Get Tom's Hardware's best news and in-depth reviews, straight to your inbox.
That may mean employment opportunities for some of the AI trainers Google let go, but it does make it clear what a shaky industry this is to get into. The big AI developers don't appear to have a coherent strategy for how to improve their products in the near term and are taking different approaches in doing so.
All the firings come in stark contrast to the major AI developer hiring sprees that companies like Meta went on earlier this year. While low-paid data annotators may have limited employment prospects at the big firms, there is less concern over hiring big names and rising stars for hundreds of millions of dollars.
Although Meta and others have seriously slowed their spending in recent weeks, massive expansion projects in the AI industry continue unabated.
Follow Tom's Hardware on Google News to get our up-to-date news, analysis, and reviews in your feeds. Make sure to click the Follow button.
Jon Martindale is a contributing writer for Tom's Hardware. For the past 20 years, he's been writing about PC components, emerging technologies, and the latest software advances. His deep and broad journalistic experience gives him unique insights into the most exciting technology trends of today and tomorrow.
-
Jabberwocky79 I have zero confidence in my ability to predict the future, but there are certainly some subtle signs that the AI bubble could be as short-lived as the dotcom bubble.Reply -
blitzkrieg316 Reply
You are the first person I've seen so far that agrees with the impending AI bubble. It's way to hard, too fast, and cannot scale unless there's a MAJOR breakthrough in power consumption reduction or a breakthrough in improving the "smartness" in a regular setting. But, when a data center takes as much power as a small city, there's going to be major problemsJabberwocky79 said:I have zero confidence in my ability to predict the future, but there are certainly some subtle signs that the AI bubble could be as short-lived as the dotcom bubble. -
redgarl Reply
That's your perception, nothing else.Jabberwocky79 said:I have zero confidence in my ability to predict the future, but there are certainly some subtle signs that the AI bubble could be as short-lived as the dotcom bubble.
If you are looking at numbers, Oracle just exploded and TSMC cannot make enough chips. People don't understand that AI is not to make money, but to save on costs due to business processes. All the governments will want their own AI to keep the control over their own data for national security reasons.
Startups are going to make it or not, but AI is not going to go away anytime soon. The next step are robotics and AGI, which defense industries are all racing towards. -
bourgeoisdude Reply
The question isn't whether or not AI is here to stay. The dot com bubble burst, but there are more for coms than ever before today. The question is whether the return that AI provides is currently overvalued. Many argue that it is and that the current trajectory is unfeasible. The AI "boom" may be over and the markets don't know it yet. Then we have the large market correction, some AI startups will fail, others will succeed. After a while we will have more AI use more prevalently, but the market will no longer overvalue it's worth. That's the argument for the 'burst'.redgarl said:That's your perception, nothing else.
If you are looking at numbers, Oracle just exploded and TSMC cannot make enough chips. People don't understand that AI is not to make money, but to save on costs due to business processes. All the governments will want their own AI to keep the control over their own data for national security reasons.
Startups are going to make it or not, but AI is not going to go away anytime soon. The next step are robotics and AGI, which defense industries are all racing towards. -
Findecanor Reply
A couple years ago I got the perception that there were a bunch of startups that made AI inference chips with analogue circuits that were supposed to make inference. use magnitudes less power than digital circuits.blitzkrieg316 said:You are the first person I've seen so far that agrees with the impending AI bubble. It's way to hard, too fast, and cannot scale unless there's a MAJOR breakthrough in power consumption reduction or a breakthrough in improving the "smartness" in a regular setting. But, when a data center takes as much power as a small city, there's going to be major problems
But training still requires digital hardware, and the AI service companies are competing with one-another on constantly getting new models out that are bigger than the ones before. That is apparently what all the power is going to.
I don't know how "AI alignment" is supposed to work. But I have a perception of how a human brain is similar to an artificial neural network. When we dream, we get free association: our minds conjure up all manner of things. But when we wake up and remember the dream, we find that many things in the dream don't make sense. I think our awake minds have a filter that keeps our subconscious thoughts in check to those that are reasonable. I think that the current crop of "AI" has only the former, and that is why they get "hallucinations". One drive factor behind companies wanting larger AI models seems to be as to avoid hallucinations by making more other valid neural pathways. But if the AI had better filters then the models wouldn't need to be as large.
Another difference between artificial neural networks and human minds is that the human minds creates models and puts new knowledge into the framework of those models, whereas NN takes raw data as coefficients.
Perhaps the way forward is something that merges the current neural-network AI with what we used to call "AI" back in the "old days" (pre 2015 ;-þ), i.e. something more similar to expert systems that actually has structured information that can be properly audited rather than only a black box of billions of coefficients. -
Eximo Everyone is scrambling to innovate. If they fail to produce meaningful/economically viable results they will sell off their hardware. So we will have a LLM hardware bubble burst as well as a large number of startups and incumbents who will lose large chunks of value.Reply -
Giroro Replyredgarl said:If you are looking at numbers, Oracle just exploded and TSMC cannot make enough chips. People don't understand that AI is not to make money, but to save on costs due to business processes. All the governments will want their own AI to keep the control over their own data for national security reasons.
Sure, some companies are taking huge amounts of money from investors right now... which is exactly what happened in the dotcom bubble before it popped.
The issue is that these companies are spending billions on hardware (and labor) right now in the hopes to save millions on labor.
Nobody knows how to make any money from AI - outside of convincing the next wave of investors to pay off the last wave of investors. Investors might stay hyped on AI forever, but without the backing of a profitable business model, they can only keep spending their money for so long until their money simply runs out. I'm not saying it's a Ponzi scheme, but it's not a sustainable industry with money flowing back and forth. All those investor dollars are definitely funneling out of companies and into the personal fortunes of a few specific individuals. It looks like those people are rushing to peak the market and cash out, not reinvest.
Investors are placing their bets based on where they think there will be a major breakthrough in either cost reduction, or ability to monetize the tech. Maybe there will be. But if it doesn't happen in the next year or two, then most of the industry is going to be out of cash.
It's not like the crypto/NFT bros have a profitable business to help them recoup their losses and fuel their enthusiasm. If AI doesn't turn a profit soon, they're just going to be bled dry and go broke. Easy come, easy go.
On the other hand, Crypto is even more of a ponzi-like scam, which is a zero-sum game that had/has no product, and no mechanisms whatsoever for the industry to generate profit.
Crypto never ran out of marks willing to gamble and lose their money, so maybe AI won't either. Heck, regular old gambling is still a thing, for some reason.
But the winners don't spend their money gambling, they spend it to buy an entire casino. -
derekullo Good news, everyone!Reply
Those asinine morons who made an AI that stole everyone's jobs were themselves fired for incompetence! -
bill001g Maybe they should have asked the AI why companies use contractors rather than employees.Reply
You get paid more money as a contractor because you know they don't give you benefits like vacation or even medical insurance if you are actually independent and do not work for a contracting firm. They also depending on the contract terms can just decide to not renew the contract for no reason at all. No severance, no unemployment
Could be they are really new and let the contracting company pay them low wages when they are billing their labor out to the actual company at a much higher rate.
People who work for gaming companies know when game releases the company only needs a fraction of the staff for ongoing support and why a lot of the positions are contractors and tend to be cut right around release date.
Why would they think developing AI is any different. -
wussupi83 Reply
Nailed it. These companies use contractors and the potential of becoming actual employees as a carrot on a stick to get long work hours from them. But then the project inevitably ends or, maybe worse, the contractors became so good at devising replicable work systems that lower skilled and wage employees can take over the job. So most workers don't end up getting any slice of the large AI pie or a stable career.bill001g said:Maybe they should have asked the AI why companies use contractors rather than employees.
You get paid more money as a contractor because you know they don't give you benefits like vacation or even medical insurance if you are actually independent and do not work for a contracting firm. They also depending on the contract terms can just decide to not renew the contract for no reason at all. No severance, no unemployment
Could be they are really new and let the contracting company pay them low wages when they are billing their labor out to the actual company at a much higher rate.
People who work for gaming companies know when game releases the company only needs a fraction of the staff for ongoing support and why a lot of the positions are contractors and tend to be cut right around release date.
Why would they think developing AI is any different.