Workers Worried About AI Job Loss, But Most Haven't Lost Their Jobs Yet

Key points

• A working paper published on SSRN in 2025 examined public attitudes toward AI-related job loss during the early generative AI period, roughly 2022 to 2024.
• Researchers found that fear of AI unemployment was widespread and crossed income levels, industries and education brackets.
• Actual measured job losses directly attributed to generative AI, the technology behind tools like ChatGPT, remained hard to isolate in the data.
• The gap between perceived threat and documented displacement is itself a significant finding, with real effects on worker behaviour and policy pressure.

What did the research actually find?

Workers across a wide range of jobs reported feeling anxious about AI, but the data on real displacement is murkier than the headlines suggest. The study, posted to SSRN, a repository for pre-publication academic research, drew on surveys and labour-market data from the period when generative AI, the kind of artificial intelligence that writes text, generates images and holds conversations, entered mainstream workplaces.

Fear was high. Documented, AI-specific job loss was not clearly measurable at scale, at least not yet.

That gap matters. When workers believe their jobs are at risk, they change behaviour: they hold off on asking for pay rises, they accept worse conditions, and they push governments to act. The researchers argue that fear itself carries economic weight, independent of whether the layoffs actually arrive.

Who felt most at risk?

Anxiety was not confined to lower-wage or purely manual work. White-collar workers, including people in writing, customer service and mid-level office roles, reported high concern. That tracks with what generative AI is actually capable of: it is better at drafting an email than fixing a pipe.

The paper notes that education level did not reliably predict lower fear. Highly educated workers in fields where AI can produce credible-sounding output, think legal drafting, financial summaries, or code, felt the pressure too.

What does this mean for ordinary workers?

If you are worried about AI and your job, the research suggests you are far from alone, and your concern is not irrational. Generative AI is genuinely capable of handling tasks that used to require a person. At the same time, the actual wave of AI-caused redundancies, at least as recorded in labour data through 2024, has not matched the level of fear.

That is not a reason to ignore the risk. It is a reason to be precise about it. Ask what specific tasks in your job AI can already do cheaply, and which parts still need human judgement, relationships or physical presence. That distinction is more useful than a blanket fear or a blanket reassurance.

Policymakers are watching the same gap. Several countries are debating new labour-protection rules tied to AI, and this kind of research, showing that fear alone changes markets, gives those arguments harder evidence to stand on.

Common questions

Is generative AI actually causing mass layoffs right now?

As of the data covered in this paper, through roughly 2024, large-scale layoffs clearly and directly caused by generative AI were not well-documented in labour statistics, though individual companies have cited AI when cutting roles.

Should I retrain for a different career because of AI?

The research does not say that. It suggests identifying which tasks within your current role are most exposed, then building skills around the parts that remain harder for AI to replicate, such as complex judgement, client relationships and hands-on work.