“AI will take my job”: why the strength and desire to learn are disappearing

You open a code editor, Figma, or Maya, but instead of curiosity, a thought arises: “Why invest another year in this profession if a neural network will soon do it for me?” The work is still there, but the future already seems less secure. And if orders truly disappear, financial losses and the feeling that experience no longer matters are added to the anxiety.

This is a generalized example of an experience, not a quote from a research participant. Scientific literature already describes its individual components: job instability, fear of skill devaluation, emotional exhaustion, and the threat to professional identity. However, the reviewed works do not provide estimates of exactly how many Ukrainian programmers, designers, or 3D animators suffer from depression due to AI.

Not every loss of motivation is depression

By WHO definition, burnout is related to chronic workplace stress that has not been successfully managed. Its signs include exhaustion, mental distance from one's job or cynicism, and reduced professional efficacy. In the ICD-11, it is a professional phenomenon, not a distinct medical condition. [8]

Depression can encompass not only work, but also relationships, leisure, and daily life. A depressed mood or loss of interest and pleasure for most of the day, nearly every day, for at least two weeks is a reason to seek an assessment. This symptom alone is not enough for self-diagnosis. [9]

Anxiety about professional future can exist without these conditions. It also deserves attention. The term "AI anxiety" in research often denotes a specific fear or a set of survey responses, rather than an independent medical diagnosis.

What science already knows about AI fear and mental health

1. The threat of job loss matters even before dismissal

A 2019 meta-analysis by Rönnblad et al. covered 16 studies of precarious employment that met quality criteria. For job insecurity, the authors obtained an overall odds ratio for worsening mental health of OR 1.52; 95% confidence interval 1.35–1.70 and rated the quality of evidence as moderate. [1]

This does not mean that "52% of people will fall ill," and it is not an assessment of risk specifically from ChatGPT. An odds ratio is not a percentage of sick people and is not the same as relative risk. The important conclusion is more modest: prolonged job uncertainty is itself associated with poorer mental health, even before dismissal occurs.

2. There is direct data on AI threat and depressive symptoms

In 2023, Xu, Xue, and Zhao surveyed 321 employees from various professions in China in two stages with an interval of about two weeks. A stronger perception of AI as a threat to professional future was associated with higher levels of depressive symptoms. The statistical model was consistent with the role of emotional exhaustion as an intermediary; the feeling of organizational support weakened some of the links. [2]

These are survey indicators, not diagnoses confirmed by a psychiatrist. Splitting the survey in time is useful, but the short interval and self-reporting do not allow for proving causality. The sample also does not represent all programmers or designers individually.

In a 2025 study, Journal of Innovation & Knowledge data from 403 employees of South Korean companies, collected at three time points,was used. The authors did not find a direct effect of AI implementation on depression in their model, but they did find an indirect link through job insecurity. [3]

This is a significant difference: psychological load may depend on how jobs change and how secure a person feels. The mere fact of an AI tool's appearance does not determine the same outcome for everyone. Three waves of observation strengthen the analysis of temporal sequence, but do not turn it into a randomized experiment.

3. Researchers are studying the fear of skill devaluation in programmers

The project by The New Developer Hicks, Lee, and Foster-Marks covered over 3000 developers in more than 12 industries. The authors described AI Skill Threat — fears that existing technical skills will quickly lose value. In the Pluralsight laboratory report, approximately 43–45% of surveyed developers demonstrated such concerns. [4]

This is not about "45% of programmers having depression." It refers to skill-related anxiety within a specific sample. The study is an observational report/preprint associated with a company that sells educational services; this is worth noting. A learning culture and a sense of belonging to a team were associated with lower threat, while a culture of intense competition was associated with higher threat.

4. AI can speed up tasks while simultaneously increasing burnout

Feng, Afroz, and Sarma investigated 442 developers at work From Gains to Strains, presented in the ICSE-SEIS 2026 materials. They combined surveys, statistical modeling, and open-ended response analysis. GenAI implementation was associated with burnout due to increased demands and organizational pressure; autonomy and learning resources acted as protective factors in the model. [5]

This raises a practical question: what does a company do with the saved time? If it is entirely converted into additional tasks, completing an individual operation faster does not guarantee an easier workday. However, the survey does not prove that every AI assistant user will burn out, nor does it measure the prevalence of clinical depression.

5. For designers, the threat concerns both skills and creative control

In a paper by Zhang et al. in Acta Psychologica 2026, responses from 382 design professionals were analyzed. Anxiety regarding job replacement, skill obsolescence, and loss of creative autonomy was associated with the intention to distance oneself from generative AI. The strongest of these associations in the model was the fear of skill obsolescence. [6]

An important limitation: the researchers measured the intention to avoid AI tools, rather than proving a loss of desire to engage in any design or learning. Furthermore, the results cannot be directly generalized to all 3D animators. However, the work shows why the pressure of constant tool updates can be accompanied by resistance, not just enthusiasm.

"Why learn if everything will be devalued?"

This thought should not be automatically labeled as laziness. Based on the cited works, an explanation can be proposed: when the future benefit of effort seems unreliable and energy is already low, learning begins to be perceived as just another demand. This is our interpretation of the combined data, not a proven universal pattern.

For a specialist, the question may not only be "will my salary remain?" but also "will I remain a person who knows how to do something valuable?". A programmer may take pride in understanding complex systems, a designer in their own solutions, and an animator in acting and precision of movement. Demonstrating rapid generation can trigger comparisons with years of training, even though it does not answer all questions about quality, revisions, and production responsibility.

Here, three different statements must be distinguished: "the tool performs part of the tasks," "clients are changing demand," and "I no longer have value." The first can be tested; the second can be verified through actual orders and job vacancies. The third is a self-assessment, not a measurement of the labor market.

Reactions also vary. Within the same professional environment, curiosity, fear, active retraining, and avoidance can coexist. It is scientifically incorrect to claim that because of AI, everyone stops wanting to work.

If AI has already taken the job: what is known

A qualitative 2025 study describes interviews with 24 IT professionals in India, who have experienced job loss or role changes related to AI automation. Themes included emotional shock, loss of professional identity, anxious reflections, social withdrawal, and a sense of betrayal by the organization. These themes were refined through a Delphi process with 20 experts. [7]

Interviews help understand experiences but do not provide the frequency of depression among all laid-off workers. Expert consensus on themes does not replace a control group and does not prove that every layoff had a single cause—AI.

For someone who has lost income, the phrase "just adapt" may be insufficient. Both practical steps regarding work and money, and the opportunity to process the loss of a familiar status, are needed. A career plan should not require immediate enthusiasm as a prerequisite for help.

What can be done without the endless chase for tools

Below are— editorial suggestions for organizing work, not a proven protocol for treating AI anxiety. They can be adapted to a person's workload and state.

1. Separate facts from predictions. Write down what has already changed: the number of orders, employer requirements, time spent on a task. Separately, note assumptions like "there will definitely be no work in a year." This is not a denial of risk, but a clarification of what can currently be influenced.
2. Choose one small experiment. For example, test one AI tool on a typical task and record the total time including corrections. For an animator—evaluate whether a specific edit remains controllable. The result may even be negative: it also reduces uncertainty.
3. Formulate a narrow learning goal. Instead of "catching up on all AI"—"mastering one operation needed for the current project." The scope should fit into available time, rather than displacing sleep and recovery.
4. Negotiate for resources, not just results. In a team, it is worth directly discussing time for learning, checking results, expected workload, and who is responsible for errors. For a freelancer, an analogue could be a clear scope of revisions and a realistic deadline.
5. Set boundaries for the information flow. If daily demonstrations of "the replacement of all professions" leave only exhaustion, you can allocate specific time for news and choose sources with methodology and limitations. This is an organizational attempt, not a proven treatment for depression.
6. Don't face the problem alone. Talking to a colleague about specific changes helps gain a different perspective. A professional community can provide support but does not replace professional help for persistent symptoms.

Responsibility cannot be shifted entirely onto the employee. WHO guidelines on mental health at work include changes in work organization, employee participation, and management training. For universal organizational interventions, the recommendation is conditional, with very low certainty of evidence: this is not a guarantee of the effect of every measure, but it is also not a reason to reduce everything to personal "resilience." [10]

When to seek help

If feelings of depression or loss of interest last for at least two weeks, or if sleep, concentration, or daily functioning worsens, consult a family doctor or a mental health professional. In case of severe deterioration, there is no need to wait two weeks. Depression has effective treatments; they should be selected with a specialist. [9]

Regular sleep and nutrition, manageable physical activity, and contact with loved ones can support your state, but they are not a substitute for treatment. If there is an immediate danger of self-harm, urgent help and the presence of another person are required. [9]

What this article does not prove

• There is no established percentage of Ukrainian IT professionals with "AI-induced depression" here.
• Self-reports, statistical models, and interviews do not allow for attributing all symptoms to a single technology.
• Data on designers is not identical to data on animators, and results from China, Korea, or India do not automatically describe Ukraine.
• For the Ukrainian reader, other possible sources of stress are also important: war, financial instability, health status, and family responsibilities. The studies considered do not isolate their contribution in an individual's case.
• Work organization tips do not guarantee job security and are not medical advice.

The fear of professional loss has studied psychological consequences. It does not imply that a person is weak or that their years of study were in vain. A useful question is: which specific requirements have changed, what support is lacking, and what the next manageable step is right now.

Further on the topic: AI, labor, and responsibility for change, why AI code requires time for verification and a AI cinema and the issue of animation controllability.