Microsoft Research Reveals Jobs Most Exposed to AI Capabilities

TL;DR: Microsoft analysed over 200,000 Copilot interactions to identify which job roles have the highest overlap with AI capabilities. Translators, historians, and writers face significant overlap (85-98%), whilst physical roles like nursing and plastering remain largely unaffected (<11%).

Microsoft has published research examining which occupations show the greatest overlap with AI capabilities, based on analysis of more than 200,000 interactions with its Copilot generative AI system. The study assessed how frequently users performed work activities similar to specific job roles and how effectively the AI tool completed these tasks.

Context and Background

The research found interpreters and translators face the highest AI overlap at 98%, meaning nearly all their work activities can be performed by large language model AI systems. Historians and mathematicians follow at 91% each, with writers at 85% and journalists at 81% overlap.

Microsoft’s analysis revealed the most common activities users seek AI assistance with involve gathering information and writing, whilst the AI itself primarily provides information and assistance, writes content, teaches, and advises. Physical occupations requiring hands-on work with people, machinery operation, or manual labour showed minimal overlap—nurses, plasterers, tyre repairers, and massage therapists all registered below 11%.

The study specifically examined large language model applications like Copilot, noting that other AI applications could affect occupations involving machinery operation, such as lorry driving, which weren’t captured in this research.

Looking Forward

Microsoft researcher Kiran Tomlinson emphasised the findings demonstrate AI’s supportive role rather than wholesale job replacement. “Our research shows that AI supports many tasks, particularly those involving research, writing and communication, but does not indicate it can fully perform any single occupation,” he explained.

The research explores which job categories can productively use AI chatbots rather than identifying roles at risk of elimination. Tomlinson suggests the technology may prove useful for many occupations, with the optimal approach involving leveraging AI’s abilities whilst complementing human strengths and respecting worker preferences.

Source Attribution:

Share this article