AI Begins to Leave Its Mark on Spanish Unemployment 2026

AI is beginning to leave a measurable mark on Spanish unemployment, with programmers and designers among the most exposed occupations. A recent working paper from the Barcelona Institute of Economics (IEB) found a small but directional link between AI exposure and joblessness.

The researchers compared 502 occupations using SEPE data from January 2021 to March 2026. For every ten additional points of AI exposure, unemployment in that occupation rose by 1.2% in the 25 months after ChatGPT launched in late 2022.

What the Data Shows

The study, a working paper published by the IEB, analyzed unemployment data from Spain's public employment service (SEPE) across 502 occupations. The period spans January 2021 to March 2026, covering the post-pandemic recovery and the arrival of generative AI.

To measure AI exposure, the team used a methodology that assesses how much each occupation's tasks can be performed by AI models. Occupations like programmers, data analysts, and designers scored high on exposure.

The key finding: a 1.2% increase in unemployment for every ten-point rise in AI exposure during the 25 months after ChatGPT's launch. Beyond that period, the percentage rises to 1.7%, though the authors caution this figure is imprecise and should be treated as indicative of a trend rather than a statistically robust result.

Why It Matters

This is one of the first empirical studies to link AI exposure to actual unemployment data at the occupation level. While the effect is small, it is statistically significant and points in a clear direction.

The findings challenge the more apocalyptic predictions of mass AI-driven job loss. Instead, they suggest a gradual impact that may be harder to detect but no less real. The authors describe it as a "crack" rather than a collapse.

For workers in high-exposure fields, the message is clear: adapt or risk becoming part of the trend. For policymakers, the data offers an early warning system to target retraining and support.

Who Is Most Affected

The study identifies programmers, data analysts, and designers as the most exposed occupations. These roles involve tasks that generative AI can assist with or automate, such as coding, data processing, and visual design.

Other high-exposure jobs include content writers, translators, and customer service representatives. In contrast, manual labor, healthcare, and education roles show lower exposure due to their reliance on physical presence and human interaction.

The researchers note that exposure does not automatically mean job loss; it means the nature of the work is changing. Some tasks are augmented, others replaced.

Key Takeaways

• For every ten-point increase in AI exposure, unemployment rose 1.2% in the 25 months after ChatGPT's launch.
• Programmers, data analysts, and designers are among the most exposed occupations.
• The effect is small but statistically significant, suggesting a gradual impact rather than a sudden shock.
• The authors caution that longer-term estimates are imprecise and should be viewed as indicative.
• The study covers 502 occupations using SEPE data from January 2021 to March 2026.

How AI Exposure Was Measured

The research team assessed AI exposure by analyzing the tasks performed in each occupation. They used a framework that evaluates how much of a job's core activities can be replicated or assisted by large language models and other AI tools.

Occupations with a high proportion of routine cognitive tasks scored higher. Those requiring physical dexterity, emotional intelligence, or complex social interaction scored lower.

This task-based approach is more granular than simply looking at job titles, allowing for a nuanced view of which jobs are truly at risk.

What to Watch Next

The authors plan to extend their analysis as more data becomes available. They caution that the current estimates are early and may change as AI adoption deepens. The background to this story is set out in Choosing Beggar Wants Designers Who Would Work For Exposure, Designer Puts Him In Place. A related thread runs through Paper Tablet Creative Gamechanger for Garden Designers. Earlier coverage in UK Youth Unemployment Crisis: How Earn & Learn Programs Help shows how the argument developed.

Key unknowns include how quickly companies will integrate AI, how workers will adapt, and whether new jobs will emerge to offset losses. The study provides a baseline for monitoring these dynamics. A closer look at the same subject appears in Extreme Job Anxiety: Navigating Long-Term Unemployment. Anyone following the wider picture can start with UK Youth Unemployment Falls Below 1 Million: A Turning Point. The same theme surfaces in Spain Opens the Black Box of Workplace Algorithms, which takes a different angle.

For now, the crack is small but visible. Whether it widens into a larger fissure depends on choices made by employers, workers, and policymakers in the coming years. Context on how this played out before sits in Over-50s Unemployment Crisis Demands Jobs for Older Workers. A companion piece, UK AI Bootcamps Tackle Youth Unemployment Crisis, approaches the same subject from another angle.