As artificial intelligence (AI) continues to advance, many workers are feeling increasingly marginalized and expressing concerns about a drop in the quality of their work. The implications of AI on employment are substantial, with an analysis by the International Monetary Fund revealing that approximately 40% of jobs worldwide could be impacted by this technology. Kristalina Georgieva, the IMF’s head, has likened this phenomenon to a “tsunami” transforming the job market.
Various workers who have trained AI models to take on some or all of their responsibilities have shared their experiences, shedding light on the emotional and professional challenges they face.
The Editor
‘I now earn less while working longer correcting the mistakes of AI editors’
Christie*, an editor specializing in academic papers for non-native English speakers, was invited to assist in training “assistant editors,” unaware that she would ultimately be replaced by AI. “There was a huge shortage of qualified editors,” explains Christie, 55, based in the UK. “I assumed they were training more people to help manage the workload.” Instead, she found herself correcting bizarre errors produced by the AI, including unnecessary punctuation and nonsensical geographic names.
Despite diligently addressing these errors, Christie observed that the mistakes persisted and sometimes even worsened. Eventually, she discovered that the so-called “assistant editors” were actually AI systems. “In a newsletter, the company admitted that all jobs would be pre-edited by this AI, leading to reduced pay for me, even though fixing AI’s mistakes takes longer than starting from scratch,” she laments.
Christie feels “devalued, betrayed, and furious” with her employer. “I prioritize work from other sources, but I’m trapped in a toxic cycle because they offer the highest volume of work. Many colleagues have quit since then,” she adds.
The Palliative Care Consultant
‘AI struggled with patients’ pronunciation’
Mark Taubert, a palliative care consultant and professor, initially welcomed the chance to work on a pilot chatbot project designed to assist patients in managing metastatic cancer and palliative care. Recording responses and feeding the system guidelines on patient interactions took “several hours,” according to Taubert, who is affiliated with Velindre University NHS Trust in Cardiff.
“Patients submitted their questions, and we incorporated informational leaflets that we had developed,” he explains. “We anticipated common queries, such as whether it’s safe to consume alcohol while on morphine.”
The chatbot’s main aim was to provide support to home patients needing urgent information, particularly regarding medication. Taubert found that while the chatbot accurately addressed about “50% of queries in a manner similar to my own responses,” it struggled with the intricacies of human pronunciation and errors.
“Patients might not use correct terminology and could mispronounce medications,” notes Taubert. “Moreover, their question structures often differ. We recognized the need for technology to adapt to human misspellings, dialects, jargon, and accents.”
After some upgrades, the system became safer, but the team also had to address sensitive topics, like how to respond if a patient inquired about ending their life. The chatbot, named Rita, was utilized for a period with numerous disclaimers before funding ceased.
“We encouraged patients to try it out but also provided links to official hospital leaflets for additional information,” Taubert clarifies. While he is open to embracing innovative technology, he remains confident that AI won’t replace his role entirely.
“Much of our work hinges on nuances in language, body language, and patient interaction,” he maintains. “In the future, AI may help relieve some of the administrative burdens, allowing me to focus more on patient care.”
The Translator
‘The overall effect is a decline in quality’
Philip*, aged 45, was tasked with training AI-driven translation engines that his superiors believed would eventually replace human translators. However, even after four years, he notes that these systems still lack reliability.
“At first, the translations were embarrassingly poor,” Philip states. “They have improved over time thanks to our corrections, yet they still produce formulaic and frequently inaccurate results, necessitating a thorough review of each translation.”
Living in New Jersey, Philip finds that AI translation does not ultimately save him time compared to direct translation. “It seems the overall effect is a decline in quality,” he asserts. “If you only need a rough translation, AI might suffice. But it is not consistently reliable, leading to potentially significant errors.”
The looming prospect of being rendered obsolete has haunted Philip for years, yet he acknowledges that the technology has not yet reached that point.
The Marketing Writer
‘Training your robot replacement feels like digging your own digital grave’
Joe*, 50, an award-winning marketing writer and content manager, recalls how his company began experimenting with AI as a productivity tool in early 2024. Despite reassurances regarding the security of his position, Joe soon found himself instrumental in developing AI workflows and best practices.
“In hindsight, I should have perceived the red flags when they had me spend the first half of 2025 creating extensive process documentation,” Joe remarks. “I mistakenly believed I would be overseeing the implementation.”
But in August 2025, just weeks after submitting his documentation, Joe was laid off. “During my exit interview, I was assured it was unrelated to my performance and was attributed to ‘market conditions,’ though the timing felt dubious,” he states. “Being instructed to train my AI replacement felt like digging my own digital grave.”
Now, Joe has learned that much of his former work has been delegated to junior staff who utilize his AI documentation for generating content. He is contemplating a shift to sales, a decision complicated by the circumstances surrounding his departure.
“I can’t say AI completely forced me out of my career, but at my age, the looming threat of AI makes me hesitant to pursue another writing position when I could face another layoff at 55,” he reflects.
The Mathematician
‘Work will look completely different in 10 years’ time, perhaps even less’
Filippo, 44, an associate professor in mathematics, has been collaborating with two startups on AI initiatives aimed at developing models that can understand and prove theorems with minimal human input. “The results are still somewhat limited, but it’s evident that these tools are rapidly improving,” shares Filippo, based in France. “With many of my colleagues exploring this AI technology, we are convinced that a mathematician’s role will evolve dramatically in the coming decade.”
“AI will likely take over mundane tasks that currently consume a significant part of our time, such as proving minor results needed for larger inquiries. However, whether mathematicians will still be essential for proving major theorems remains uncertain,” he notes.
Despite his explorations in AI, Filippo does not feel immediate concerns for his job stability. “Given my work at a public institution, where a considerable amount of my time is devoted to teaching, and since these AI tools are yet to reach professional research standards, I feel secure for now,” he explains. “However, I would likely have a different outlook if I were a recent PhD graduate.”
*Names have been changed
