What Jobs Will AI Replace First? What the 2026 Evidence Shows

You have probably seen the lists: "10 jobs AI will replace first." Most trace back to a headline, not a dataset. This guide goes to the primary sources instead, including Microsoft Research, Stanford and ADP, the World Economic Forum, Goldman Sachs, the U.S. Bureau of Labor Statistics and the IMF. It separates what those sources measure (exposure) from what they cannot promise (replacement), and it ends with steps you can take this week.

Last reviewed October 12, 2026. Every figure below carries its source and publication year, because these numbers move.

Quick Answer: The evidence points to information-heavy work as the most exposed: translation, writing, customer service, sales communication, clerical and data-entry roles. Hands-on, in-person and high-judgment jobs are the least exposed. But "exposed" is not "replaced." The clearest measured effect so far is not mass layoffs. It is weaker hiring of early-career workers in AI-exposed occupations (Stanford/ADP, 2026).

What Jobs Will AI Replace First?

Based on current evidence, the roles under the most pressure are those built mostly around producing, processing or communicating information: translators, writers, customer service and sales representatives, clerks and data-entry staff. No study can name the first jobs to vanish, but several independent sources point at the same cluster of work.

A pattern runs through the table. The 2025 Microsoft Research study analyzed 200,000 anonymized Bing Copilot conversations and found that the most common and most successful AI-assisted activities are information work: creating, processing and communicating information. Roles dense in those tasks score highest.

Exposure Is Not Replacement: What Do the Studies Actually Say?

Exposure means AI can assist with or perform some tasks in a job. Replacement means the job disappears. The studies themselves insist on the difference, and most headlines ignore it.

Microsoft Research states that it would be a mistake to conclude that high-applicability occupations will be automated and lose jobs or wages, because downstream effects are hard to predict. Its example is the ATM: the machine automated a core teller task, yet the number of bank teller jobs increased as banks opened more branches at lower cost and tellers shifted toward relationship work. The authors add that AI's effect on employment and wages "will depend on hard-to-predict business decisions."

Three other findings keep the picture honest:

• Stanford/ADP (August 2026 update): the authors find no evidence of widespread, economy-wide job displacement from AI. The gap shows up in hiring of 22-25 year olds, not in separations, and it is concentrated in occupations where AI usage substitutes for human tasks. Where AI complements workers, employment is flat or rising. The authors call their results early, descriptive indicators rather than causal estimates.
• Goldman Sachs Research (2025): its baseline assumes 6-7% of US workers could be displaced if AI is widely adopted, with a range of 3% to 14% under different assumptions. If only current use cases were expanded across the economy, it estimates 2.5% of US employment would be at risk. It also notes that displacement from past technology shifts tended to fade within about two years.
• IMF (2024): almost 40% of global employment is exposed to AI, about 60% in advanced economies, 40% in emerging markets and 26% in low-income countries. Roughly half of exposed jobs may benefit from AI integration, while in the other half AI may perform key tasks, which could lower labor demand.

On the global picture, the World Economic Forum's Future of Jobs Report 2025 (based on over 1,000 employers representing more than 14 million workers) projects that structural change from 2025 to 2030 will create 170 million jobs and displace 92 million, a net gain of 78 million. About 40% of employers anticipate reducing their workforce where AI can automate tasks.

Which Jobs Are Safe From AI?

No job is completely safe, but work that is physical, in-person or built on high-stakes judgment shows the lowest exposure. The same sources that flag risk also identify who is least affected.

Two cautions apply. First, "low exposure" is a statement about today's tools and today's tasks, and Microsoft Research notes its scores reflect only one slice of the AI market. Second, a low score does not guarantee good pay or demand. It means AI overlaps less with the daily tasks.

Software development is the instructive edge case. Computer occupations score high on AI applicability, and Stanford/ADP's early-career gap is large in exposed fields, yet BLS still projects 10% growth for the occupation group. That tension is why the safest reading is task-level, not title-level.

Will AI Take My Job?

Probably not wholesale, but parts of your job will likely change, and entry-level workers face the most measurable pressure today. The evidence supports "your tasks will shift" far more strongly than "your job will vanish."

The Anthropic Economic Index (June 2026) adds the worker's view. In its first survey, early-career workers reported that AI can already do the highest share of their work and expressed the most concern about job loss. About 10% of respondents rated losing their own job in the next year as likely or very likely, and the survey itself is not representative of the general population. Many respondents named judgment and management as capabilities AI lacks.

Meanwhile, Stanford/ADP finds that experienced workers show no comparable employment gap. That points to a practical reading: the early rungs of exposed career ladders are narrowing, so how you build experience matters. Our guide to the job search playbook for the class of 2026 covers that angle for new graduates.

What Skills Should I Learn to AI-Proof My Career?

Learn to work with AI and strengthen the skills it handles poorly: judgment, creative problem-solving, relationships and adaptability. The WEF's 2025 report ranks AI and big data, networks and cybersecurity, and technology literacy as the three fastest-growing skills through 2030, followed by creative thinking, resilience, flexibility and agility, and curiosity and lifelong learning.

There is also a pay signal. An IMF analysis (January 2026) found that job postings requiring new skills pay roughly 3% more in the UK and US. For more on how that plays out in postings, see our breakdown of what job postings reveal about the AI skills salary premium, and our guide to the five skills that hold their value as AI spreads.

A practical skills stack:

1. AI fluency: use AI on real work tasks weekly, learn where it fails, and be the person who checks its output.
2. Domain depth: the Stanford/ADP paper suggests occupations involving tacit knowledge see faster employment growth for experienced workers, so deepen the knowledge that is hard to write down.
3. Human-facing skills: negotiation, client management, mentoring, and accountable decision-making.
4. Adaptability: the WEF finds 39% of workers' existing skill sets are expected to be transformed or become outdated by 2030, and half of workers have already completed training or upskilling.

How Can I Check My Own Role's AI Exposure?

List your tasks, then sort them by how well AI handles each one. You do not need a proprietary tool: public data gives you a starting point.

1. Break your job into 10-15 recurring tasks. The BLS Occupational Outlook Handbook and O*NET OnLine list standard duties for your occupation.
2. Tag each task: information work (drafting, summarizing, answering routine questions, data entry) versus judgment, physical or relationship work.
3. Look up your occupation in the public data. Microsoft Research published its occupation-level results on GitHub, and the BLS handbook shows 2025-2035 projections.
4. Try the AI tools on your top five tasks. Note what is faster, what needs review, and what fails.
5. Reposition around the gap. Move your time toward the tasks AI cannot own (client judgment, accountability, coordination) and become the person who directs the AI on the rest.

If you are job hunting, our guide on how to job search in the AI era shows how to present AI skills credibly.

The Bottom Line

The honest answer to "what jobs will AI replace first" is that the evidence tracks tasks, not job titles. Information-heavy, entry-level work is where hiring has weakened most so far, while the research agrees that broad displacement has not shown up yet and that outcomes depend on employer choices. Audit your tasks, build AI fluency, and invest in judgment and relationships. Then re-check the data: most of these sources update at least yearly.

Sources and dates: Microsoft Research (2025, v6 December 2025); Stanford Digital Economy Lab with ADP (August 2026 update); World Economic Forum (January 2025); Goldman Sachs Research (August 2025); BLS Occupational Outlook Handbook (2025-2035 projections, 2026 edition); IMF (January 2024 and January 2026); Anthropic Economic Index (June 2026). Full links are in the sources file.