Contrary to the widespread assumption that AI mainly takes over routine work, more demanding tasks are increasingly being delegated to AI in the workplace – including activities that require a higher level of skill. This is shown by a new Anthropic study, which analysed millions of usage records from Claude. According to the findings, employees in the AI era are increasingly left with routine tasks, while more complex activities are shifted to AI systems.
One example cited in the study is work in travel agencies. Here, AI takes on tasks such as the “planning, description, organisation and sale of package holidays”. Activities such as “printing or requesting tickets” and “collecting payments for transport and accommodation”, by contrast, remain with employees.
According to Anthropic, this form of “deskilling” is particularly concerning because AI systems perform more demanding tasks less reliably and therefore achieve lower success rates. On Claude.ai, tasks requiring less than a high school education have a success rate of 70%, while tasks at college level achieve only 66%. Employees therefore face a trade-off between lower accuracy and greater time savings – and often opt for the latter.
However, not all occupations are affected by deskilling to the same extent. The study cites property management as a counterexample. In this field, it is mainly administrative tasks that are delegated, while more complex activities – such as structuring financing and stakeholder management – continue to be handled by professionals.
By Maria Kleiner
Sources: Heise online, 16 January 2026, by Malte Kirchner; t3n digital pioneers, 16 January 2026, by the editorial team
![AI use at work: deskilling vs efficiency gains 1 © [2025] [𝘒𝘐-𝘎𝘦𝘯.] 𝘔𝘢𝘳𝘤𝘦𝘭 𝘖𝘩𝘳𝘦𝘯𝘴𝘤𝘩𝘢𝘭𝘭. 𝘉𝘢𝘴𝘪𝘦𝘳𝘦𝘯𝘥 𝘢𝘶𝘧 𝘞𝘦𝘳𝘬𝘦𝘯 𝘷𝘰𝘯 𝘈𝘯𝘥𝘳𝘦𝘢𝘴 𝘖𝘩𝘳𝘦𝘯𝘴𝘤𝘩𝘢𝘭𝘭.](https://research-ki.de/wp-content/uploads/2026/02/an_owl_and_a_ai_robot_face_looking_at_eachother_ink_dr_97be154f-096b-4b60-a995-3533b440c3e8_1.png)