Task Expansion with Generative AI: Experimental Evidence on Middle-Skilled, Early-Career Workers

Hanna Brosch, Christina Langer, Philipp Lergetporer, Sandra Mueller, Harald Ulrich Pfeifer, Nele Tschoepe · CESifo · 7 Oct 2026

Read original sourceOpen access

Where this sits in the debate

Will AI actually boost productivity and growth? · its finding supports the claim (why studies differ)
Will AI destroy more jobs than it creates?

Direction of the finding as read from the abstract by an AI model (AI-written).

Key takeaway AI-written

A field experiment randomly gave 673 final-year IT apprentices at twelve German vocational schools access to generative AI for occupation-specific tasks. AI raised productivity across all task types, most where unaided performance was lowest, and by 26-31 percentage points on tasks beyond the apprentices' formal training, without reducing immediate task comprehension.

The paper in three steps AI-written

Written by an AI agent from the paper's abstract, which is the primary source and is quoted below.

The question

Does generative AI let middle-skilled, early-career workers successfully perform tasks they could not do before?

The approach

A field experiment with 673 final-year IT apprentices at twelve German vocational schools, randomly assigning AI access for occupation-specific tasks both within and beyond their curriculum.

The findings

AI raised productivity across all task types, with larger gains where unaided performance was lowest, and by 26-31 percentage points on tasks beyond formal training. Immediate task comprehension did not fall, and effects were larger for apprentices with above-median AI literacy.

How to read it: experiment. Measures a causal effect in a controlled or randomized setting — strong internal validity; ask how far it generalizes.

Limitations and counterpoints AI-written

The setting is German IT apprentices completing set tasks, so results may not carry over to other occupations, experience levels or real workplace output over time.

Abstract

Does generative artificial intelligence (AI) expand the set of tasks that middle-skilled, early-career workers can successfully perform? In a field experiment with 673 final-year IT apprentices at twelve German vocational schools, we randomly assign access to AI for completing occupation-specific tasks. To identify task expansion, we study tasks both within the apprenticeship curriculum and deliberately beyond it. AI raises productivity substantially across all task types, with gains tending to be larger on tasks where performance without AI is lowest. Crucially, productivity increases by 26–31 percentage points on tasks beyond apprentices’ formal training, providing direct evidence of AI-enabled task expansion. These productivity gains do not come at the cost of reduced immediate task comprehension. Treatment effects are generally larger among apprentices with above-median AI literacy.

About this publication

Academic journal. Published in an academic journal — typically peer-reviewed before publication, though indexing alone doesn't guarantee review depth.

Published: 2026 (recently)
Citations: 0 (OpenAlex)

0 citations recorded by OpenAlex — too few 2026 papers in this collection for a fair comparison.