Worried about AI taking your job? This chart from Anthropic reveals the most at-risk roles

Some jobs could be in the danger zone

Concerns about artificial intelligence replacing human workers have been growing for years. Now, economists working with Anthropic say they have created a new way to track which jobs may face the greatest disruption from AI tools like large language models.

Their research combines real-world data from how people use Claude with an analysis of tasks AI could theoretically perform. The result is a chart that highlights which occupations may face the most exposure as AI capabilities continue expanding across industries.

The jobs that may face the most AI exposure

Using their new measurement system, Anthropic economists identified the occupations with the highest observed exposure to AI. These are roles where many day-to-day tasks can already be assisted or partially handled by large language models.

The most exposed occupations in Anthropic’s ranking include computer programmers, customer service representatives, data entry keyers, medical record specialists, and the combined category of market research analysts and marketing specialists. These jobs center on digital information work, where language models are currently most useful.

Why programmers top the list of AI vulnerable roles

Computer programmers appear at the very top of the exposure rankings. That might surprise some people because tech jobs are often seen as the safest careers in the digital age.

But tools like Claude are widely used for coding tasks such as generating scripts, debugging problems, and writing software functions. Researchers say that heavy AI use in coding work is one reason programming roles appear especially exposed in their analysis.

A new way economists are measuring AI job disruption

The research introduces a new measurement called Observed Exposure. Instead of guessing how AI might affect jobs someday, the method uses real usage data from Anthropic’s Claude model combined with an analysis of job tasks.

This approach helps researchers understand how AI tools are actually being used across different types of work today. By studying real behavior rather than predictions alone, economists hope to track labor market changes more accurately.

The data behind Anthropic’s economic tracking project

The study draws on Anthropic’s Economic Index work, which analyzes how people use Claude in real-world settings and maps that activity to occupational tasks. It combines observed AI usage with labor-market data to examine where language models are already being used in work-related tasks.

Anthropic has also published separate geography-focused research on Claude usage across U.S. states and the District of Columbia. Together, those projects aim to give economists an earlier and more empirical view of how AI may be reshaping work.

AI has not yet triggered major job losses in these fields

Despite high exposure levels, Anthropic’s researchers say they do not yet see evidence of major unemployment spikes in the occupations most exposed to AI.

They do, however, report tentative signs that hiring has slowed for younger workers entering some of these highly exposed fields. In particular, the economists saw evidence suggesting that the hiring of younger workers in some highly exposed fields has slowed in recent years.

Little-known fact: Anthropic analysis of millions of Claude conversations found that about 36% of occupations use AI for at least a quarter of their tasks, showing how widely the technology is already spreading in the workplace.

AI still cannot perform many real world job tasks

Even the most advanced language models have clear limitations. Many types of work involve physical activity, real-world judgment, or legal authority that AI systems simply cannot replicate.

Examples include agricultural tasks such as pruning trees or operating farm machinery. Legal work, such as representing a client in court, also remains outside the reach of current AI technology.

How far AI still has to go before replacing tasks

The researchers say current AI usage is still far below the theoretical limits of what large language models could eventually handle in the workplace.

For example, Claude currently covers only about 33 percent of tasks within the computer and mathematics category. That gap shows there is still a large difference between AI’s potential capabilities and how it is actually used today.

Dario Amodei’s warning about white collar jobs

Anthropic CEO Dario Amodei has been one of the most outspoken voices warning about potential job disruption caused by artificial intelligence systems.

He has previously said AI could replace up to half of entry-level white-collar jobs within the next one to five years. While some industry leaders disagree with that timeline, the debate highlights growing concern about automation.

Why entry level tech roles may be the first to change

Recent research suggests younger workers may be feeling some of the earliest labor-market effects in AI-exposed occupations. Anthropic cites evidence that hiring into highly exposed roles has slowed for workers ages 22 to 25, even though broad unemployment spikes have not yet appeared.

That pattern fits concerns about early-career jobs built around routine digital tasks such as coding, document work, and analysis. The long-term impact remains uncertain, but researchers say hiring trends for younger workers are worth watching closely.

Some hands on jobs may remain safer from AI

The study also highlights professions with much lower exposure to AI. These jobs tend to involve physical work or real-world environments that digital systems cannot easily navigate.

Examples include cooks, motorcycle mechanics, lifeguards, bartenders, and dishwashers. Work that involves moving objects, interacting with people directly, or operating equipment may be harder for AI to automate.

Little-known fact: Researchers at OpenAI found that around 80% of U.S. workers could see at least 10% of their job tasks affected by large language models.

Why AI job predictions have been wrong before

Predictions about AI replacing workers have often moved faster than reality. In 2016, Geoffrey Hinton said radiologists could be overtaken quickly, but the profession remains active and in demand years later.

Hinton later acknowledged that he had spoken too broadly and got the timing wrong. The episode is a reminder that AI can change how work is done without eliminating an entire profession on the schedule, early forecasts suggest.

If you want a glimpse at the hardware powering the next wave of AI, check out how OpenAI launched a model built on Cerebras chip technology.

AI disruption may affect workers very differently

Researchers also found that workers in highly exposed professions tend to share several demographic patterns. These jobs are more likely to be held by people who are older, highly educated, and earning higher salaries.

The economists also noted that women appear more frequently in several of the most exposed professions. This suggests that AI-driven workplace changes may not affect every group of workers equally, and certain demographics could experience stronger shifts as companies adopt more automation tools.

The tech is evolving faster than you think. Take a look at how OpenAI is introducing age prediction tech to ChatGPT.

What do you think about which jobs could face the biggest AI disruption in the coming years? Share your thoughts.

This slideshow was made with AI assistance and human editing.

Don’t forget to follow us for more exclusive content on MSN.

Read More From This Brand:

• Bill Gates introduces OpenAI alliance focused on health systems
• OpenAI’s ChatGPT might lose safety guardrails after long-term use
• OpenAI funds Sam Altman’s neural interface company, Merge Labs