Skip to content
Neural Daily

BusinessNews

Hungary's AI exposure map: two thirds of workers affected

🇭🇺 Olvasd el magyarul →

BeHive Consulting and AI Budapest have built the first AI exposure map covering Hungary's entire workforce. Across 462 occupations and 4.67 million workers, AI is set to augment the daily tasks of 54% and transform those of 9% over the next two to three years.

· 3 min read

Felülnézeti kép: pénzügyi kimutatások, számológép, bankjegyek és laptop egy íróasztalon
Image: Pexels / Tima Miroshnichenko · Pexels · Source

What was built

The joint study by AI Budapest and BeHive Consulting covers all 462 four-digit FEOR-08 occupations and 4,668,600 workers (aimunkaterkep.com). It does not measure whether jobs disappear but how much AI could rewrite the mix of daily tasks over the next two to three years. The interactive heatmap is free and needs no registration.

Three groups

  • Transforming roles, 9% (about 414,000 people, score 8–10): tasks change substantially; worth planning reskilling paths.
  • Augmentable roles, 54% (about 2.51 million, score 4–7): AI takes over some tasks, a direct productivity opportunity.
  • Structurally stable roles, 37% (about 1.75 million, score 1–3): physical, care and skilled trade work such as cleaners, truck drivers and welders.

The headcount-weighted average AI impact is 4.6 out of 10. The top two groups together make up the roughly 64% figure quoted in the Hungarian press (Portfolio).

Who is most affected

Thirty-two occupations fall into the transforming band, and the top of the list is office work rather than tech: general office administrators (60,845 people, 8/10), secretaries (31,266, 8/10), accountants (24,547, 8/10) and tax office clerks (24,436, 8/10). Data entry clerks and coders got the only 10/10 (21,842 people).

Software developers scored 9/10 (21,039 people) and systems analysts 8/10. The researchers stress this does not mean job losses: demand for software may even grow as each developer gets more productive. According to Portfolio, in this group it is mainly entry-level developers whose work changes.

How it was measured, and its limits

The method is a Hungarian adaptation of Andrej Karpathy's open-source framework, originally built on US BLS data. Headcount and wages come from KSH STADAT, outlooks from Cedefop's 2022–2035 skills forecast. An LLM scored AI impact per occupation from FEOR task descriptions, with a written rationale for each score. The source code is public.

Two important limits: the researchers themselves call the scores rough estimates, not statistical forecasts. And the map covers software AI only, not robotics or industrial automation. In other words, an LLM is judging how far LLMs can take over tasks, which deserves some caution.

What it means in practice

For anyone running AI adoption in a company it is a useful starting point: the high-headcount administrative, accounting and clerical roles are where most people's work will change, not the developer teams. The heatmap shows impact, outlook, hiring demand and wages per occupation, so it can guide reskilling plans or the choice of a pilot project. Separate summaries for SMEs and large companies are available on request.

Key takeaways

  • 462 occupations, 4.67 million workers, a two-to-three-year horizon.
  • 54% augmentable, 9% transforming, 37% stable.
  • Office administration and accounting are most affected.
  • An LLM-based estimate, not a forecast.

Related articles