Workers grappling with the rapid growth of artificial intelligence say they feel “devalued” by it and warn of a decline in the quality of their jobs.
A recent analysis by the International Monetary Fund found that AI will impact around 40% of jobs worldwide. “This is like a tsunami hitting the labor market,” said Kristalina Georgieva, head of the association.
Employees who have trained AI models to replace some or all of their roles speak to the Guardian about their experiences.
editor
“My income has decreased because I work long hours to correct mistakes made by AI editors.”
Christie* edits papers for academics for whom English is a second language. She is asked to participate in a project to train new “associate editors” without knowing that it is an AI program. As a result, my salary has been reduced.
“I thought they were better trained because there was a huge shortage of qualified editors.” [people] To take some of the burden off,” says Christie, 55, who lives in the UK. “Then they instructed me to correct the deputy editor’s mistakes. However, the new editor was making strange mistakes, such as inserting unnecessary punctuation and changing country names to meaningless ones.”
“I have pointed out these errors with great care and respect,” Christie said.
However, the errors continued to occur and “in some cases got worse.” Then, a few months later, she found out who the “editor” was.
“In the newsletter, the company acknowledged that these deputy editors are actually AI,” Christie says. “from now on, All work is pre-edited by AI, so you get a reduced fee, so you have less income to fix the AI’s mistakes, and it takes more time than editing from scratch.
“There is a collective feeling within the company that we have to implement AI.”
Christie said she feels “devalued, betrayed and furious by this company.”
“I prioritize work from other sources, but I get the most work from other sources, and I have to eat and pay rent, so I get stuck in this toxic cycle. But a lot of people quit,” she added.
Palliative care consultant
“The AI had difficulty understanding the patient’s pronunciation.”
Palliative care consultant and professor Mark Taubert said he was excited to work on a pilot chatbot project to explore how technology can help patients navigate the complexities of metastatic cancer and palliative care.
Ms Taubert, 51, who works at Velindre University NHS Trust in Cardiff, entered guidelines into her computer that were recorded for “several hours” for the chatbot and generally informed how to speak to patients.
“We asked patients to write down all their questions and added a pre-written and agreed-upon patient information leaflet,” he says. “We also considered questions that palliative care outpatients and inpatients might ask, such as, ‘Can I drink alcohol while taking morphine?’”
The chatbot was primarily aimed at homebound patients who had questions about medications, etc. outside of business hours.
Taubert said the chatbot was “about 50% accurate in the way I responded,” but struggled with the vagaries of human pronunciation and human error.
“Patients don’t always use perfect English and sometimes use the names of drugs incorrectly. For example, they might say ‘morphium’ instead of morphine,” he says. “People were also structuring questions very differently. We recognized the need for technology that learns human spelling mistakes, dialects, jargon, variations, and accents.
“Subsequent adaptations have made the system more secure, but we also had to consider how the machine would react if a patient entered more thorny questions, such as how to end their life.”
The chatbot, called Rita, had been in use for a while “with a lot of caveats and caveats” before funding ended, Taubert said.
“We say, ‘Try it if you like,’ but we also include links to hospital information leaflets for each area,” he added.
Although Taubert is open to “embracing new technology,” he doesn’t feel his role is threatened by AI.
“A lot of what we do depends on the nuances of our words, our body language, our facial expressions, and whether we’re in the room,” he says. “In the coming months and years, a system like this will eliminate the whole administrative task, allow me to actually talk to patients more, and probably enhance my work week.”
translator
“The overall effect is a decline in quality.”
Philip*, 45, has had to train an AI-based translation engine that his boss “wants to replace because it’s cheaper,” but says it’s still unreliable four years later.
“At first, the results were necessarily laughable,” he says. “But as we modified the program, it got better. But even after all these years, it still tends to produce boilerplate results and is still unreliable and not accurate enough. That’s why we have to review the AI-generated translations word-for-word and correct them if necessary.”
Phillip, who lives in New Jersey, speaks from his own experience. “It doesn’t save you any time compared to directly translating the material yourself. I think the overall effect is a loss of quality.” If you only need a rough idea of what’s being said, AI is generally fine. But it’s not always reliable, and that’s the problem. Because sometimes you’ll come across things that are completely wrong. ”
He said the moment when he would no longer be needed in his current role “has been looming over our heads for years, but we’re not there yet.”
marketing writer
“Training a robot replacement feels like digging your own digital grave.”
Joe*, 50, an award-winning marketing writer and content manager, said the company he worked for started considering AI as a productivity tool in early 2024, but he was confident his job was secure.
“I should have seen the writing on the wall when I was tasked with spending the first six months of 2025 building extensive ‘AI process workflows’ and ‘best practices documentation.’ In my naivety, I thought I would be managing this system and would be asked to oversee these processes.”
But in August 2025, two weeks after submitting the best practices document, Joe was fired.
“During my exit interview, I was told that it had nothing to do with my work or performance. They blamed it on ‘market conditions,’ and while some of that was definitely true, the timing was certainly questionable,” says Joe, who lives in Milwaukee. “Working for this company and being asked to train robot replacements feels like digging my own digital grave.”
Joe has been told that much of his previous work has been delegated to younger employees.
“They are following “My AI documentation simply prompts the AI client to produce the work I used to do,” he says.
Joe is currently considering a career change into sales, but says it hasn’t been easy.
“I wouldn’t necessarily say that AI has 100% thrown me out of my career path, but as I turn 50 and with the ever-looming threat of AI, I wonder if I’ll get fired again when I’m 55, when I could have taken another writing job.”
mathematician
“In 10 years, jobs will be very different. Maybe even less.”
Filippo, 44, an associate professor of mathematics, is collaborating with two startups on AI projects.
They develop models that reason mathematics and prove theorems with little human input, and use proof assistant software Lean to verify inputs.
“Three months have passed and the results are still somewhat limited, but it is clear that these tools are becoming more powerful and more efficient every day,” says Filippo, who lives and works in France. “With most of my colleagues experimenting with this AI technology, we are confident that the work of mathematicians will look very different in 10 years, or even less.
“AI will be able to replace us with mundane tasks that take up a lot of our time, such as proving small ancillary results needed for larger goals. Whether mathematicians are still needed to prove these larger goals is debatable.”
Filippo, who works at a university, said she doesn’t feel her role will become obsolete in the near future.
“Given that I work in the public sector and spend a lot of time teaching, and these AI tools have not yet reached the level of professional research, I don’t feel any pressure or concern about my work,” he says. “But if I had just finished my PhD at 25, I would have had a very different perspective.”
*Name has been changed
