From engineers to financial analysts: Anthropic study reveals jobs that AI has or will replace in future

anthropic

• AI could automate tasks in jobs like programming and data entry
• Study finds limited evidence of AI causing job losses so far
• Physical or in-person roles less exposed to automation

Did our AI summary help?

Artificial intelligence could increasingly take over several workplace tasks across white-collar professions, according to a new study by AI company Anthropic. The research examines how closely current AI capabilities match the activities performed in different occupations and highlights roles where automation could become common.

Anthropic said the research is part of an “early warning system” designed to track how AI may influence the labour market over time. By comparing AI capabilities with task requirements across professions, the study identifies jobs where a large share of daily work could potentially be handled by AI systems.

While the findings suggest AI could automate or assist with several tasks, the company said there is currently limited evidence that the technology has directly reduced employment levels.

Jobs most exposed to AI automation

The analysis found that several occupations involve tasks that align closely with what modern AI systems can perform.

According to the study, the jobs with the highest exposure to AI-driven automation include:

Computer programmers — 75 percent of tasks

Customer service representatives — 70 percent

Data entry keyers — 67 percent

Medical records specialists — 67 percent

Market research analysts and marketing specialists — 65 percent

Sales representatives — 63 percent

Financial and investment analysts — 57 percent

Software quality assurance analysts — 52 percent

Information security analysts — 49 percent

Computer user support specialists — 47 percent

Anthropic clarified that exposure does not mean entire jobs will disappear. Instead, it indicates that a significant portion of tasks within these roles could be automated or supported by AI tools.

For example, AI systems may help programmers generate code snippets, assist analysts with data processing, or respond to routine customer service queries.

Despite concerns about large-scale job losses, the research found little direct evidence that artificial intelligence has already reduced employment levels across industries.

However, researchers observed early signals that hiring could slow in occupations where AI tools are becoming widely used. Jobs with higher AI exposure may also see slower employment growth over the coming decade.

Anthropic added that the technology is more likely to transform how work is performed rather than eliminate entire professions.

Jobs less likely to be replaced by AI

The study also found that roles requiring physical skills or in-person interaction remain less vulnerable to automation.

Examples include cooks, bartenders, lifeguards, motorcycle mechanics and groundskeepers. These jobs depend on physical activity, real-world judgement and human interaction that current AI systems cannot easily replicate.

Anthropic researchers said understanding the limits of AI will be important for workers, businesses and policymakers as the technology continues to evolve.