Robots Are Better Than People home
Subscribe

Will AI Replace IT professionals?

AI now writes a large share of new code at some of the biggest tech firms, and several software companies have cut staff while citing AI. Official projections still show computer jobs growing, though early-career workers are taking the first hit.

Our readGrowing overall; support roles exposedOur read is that AI is absorbing routine coding and support tasks while total computer employment keeps growing, but entry-level and support roles face real pressure.

The morning starts with a queue. A user is locked out, a deployment failed overnight, a pull request needs review before noon, and a network link keeps dropping between two offices. Somewhere in that list, an AI assistant has already drafted a fix, suggested a log query, or written half the code waiting for review.

That is the real question for IT professionals. AI is not arriving later. It is already in the queue. What matters is which parts of the work it takes, and who still gets paid to do the rest.

The headline numbers point in two directions. The U.S. Bureau of Labor Statistics projects computer occupations will grow from 5,036,200 jobs in 2025 to 5,323,800 in 2035, a +5.7% change against 3.5% for all occupations, with 280,000 openings a year on average and a $109,470 median annual wage. At the same time, Stanford Digital Economy Lab researchers found employment for software developers aged 22-25 declined nearly 20% compared to its peak.

What to do with this, depending on where you stand

  • Students: The field is still projected to grow, but the entry door is narrower. The Stanford Canaries Dashboard shows large declines for early-career workers (22-25) and expansion for older age groups. Plan to show judgment and review skills, not only the ability to produce code.
  • Mid-career workers: Experience is currently your shield. The same Stanford data shows modest declines for the next-youngest groups and growth beyond them. Move toward work that involves owning systems, approving changes and building AI-based tools.
  • Employers and clients: Do not assume AI tools speed everyone up. A METR randomized trial found experienced developers took 19% longer to complete issues when allowed to use AI. Measure results in your own codebase before cutting staff.

The IT job, task by task

TaskWhat AI does todayWhat still needs a person
Writing new codeGenerates large shares of new code at Google, Microsoft and CoinbaseApproving it. Google's figure counts code approved by an engineer
Debugging and modifying softwareAmong the most common tasks people bring to Claude, per AnthropicDeciding which fix fits the system and its users
Network troubleshootingAnswers troubleshooting queries in chat toolsPhysical checks, vendor escalation, change approval
User support ticketsChatbots handle troubleshooting, per a summary of BLS projectionsUnusual problems and frustrated users
Repetitive programmingCompanies are expected to automate it, per BLSArchitecture and requirements
Building and maintaining AI systemsLittle on its ownDevelopers to build AI-based business solutions, per BLS

Where AI is already writing code and cutting IT jobs

Big tech firms now report AI-written code as a share of their output. Google's own figure moved from more than a quarter of new code in October 2024 to 50 percent in autumn 2025, and then to 75 percent for new code that is AI-generated and approved by an engineer.

Microsoft CEO Satya Nadella told a Meta conference that 20% to 30% of code in the company's repositories was written by AI, with more progress in Python than in some other languages. An industry roundup reports roughly 40% of daily code at Coinbase is AI-generated.

Usage data backs this up. Anthropic found 79% of conversations on Claude Code were automation, where AI directly performs tasks, against only 49% on Claude.ai.

Layoffs have followed in some companies:

  • Amazon cut about 14,000 corporate roles in October 2025 and a further 16,000 in January 2026, with the second wave tied to AI adoption.
  • GitLab cut around 14% of its workforce, about 350 people, in June 2026 as it leans on AI agents.
  • The same tracker lists 21,000 workers at Oracle based on its annual SEC filing, and a 15% cut at Mews in July 2026, which Mews called a "proactive decision".

Across the whole U.S. economy, not just IT, AI was cited in 87,714 announced job cuts through May 2026, or 22% of all layoffs. Those are employer-stated reasons, not a direct count of jobs AI replaced.

The messy middle: speed claims, measurement and lawsuits

Company code percentages are self-reported. TechCrunch noted it is unclear how Microsoft and Google measure what counts as AI-generated.

Productivity is contested. The METR trial used 16 experienced open-source developers and 246 tasks in mature repositories. METR itself cautioned against overgeneralizing, and a later write-up says METR called follow-up data unreliable because 30-50% of developers withheld tasks they did not want to do without AI.

Legal questions remain open. On September 16, 2026, the Ninth Circuit affirmed dismissal of a DMCA claim in Doe v. GitHub over Copilot. The court expressed no view on whether Copilot output infringes copyright, and two breach-of-contract claims were not resolved by that decision.

Exposure is not destiny either. BLS rates software developers "Very high" on its AI exposure scale, but BLS states a high exposure category is not a forecast of job loss.

Skills that gain value as AI writes more code

  • Building and running AI systems. BLS expects AI adoption to fuel strong job growth in computer and mathematical occupations.
  • Code review and approval. Google's 75 percent figure only counts code an engineer approved. Someone accountable still signs off.
  • Software development over pure programming. Software developers are projected to grow 10%, adding 174,700 jobs, while BLS expects firms to automate repetitive programming tasks.
  • Licensing and IP awareness. The Copilot litigation shows teams need people who understand code provenance.

Note the softening signal: BLS no longer lists AI research as a demand factor for software developers, after listing it the year before, when its earlier projection was 15.8 percent growth.

Our verdict: AI is taking tasks, not the whole IT department

We think AI will keep absorbing code writing, debugging drafts and first-line support. Our read of the evidence is that total IT employment keeps growing, while support roles, routine programming and entry-level developers face the sharpest squeeze.

The bottom line: IT work is being reshaped around reviewing, owning and building AI systems. If you are early in your career, build proof that you can judge and ship code, not just type it.

Questions people ask

Will AI replace IT professionals?

BLS projects computer occupations to grow from 5,036,200 jobs in 2025 to 5,323,800 in 2035. Our read is that AI replaces tasks such as routine coding and first-line support rather than the whole profession.

How much code does AI write today?

Google says 75 percent of new code is AI-generated and approved by an engineer, and Microsoft's CEO said 20% to 30% of its code is written by AI. TechCrunch noted it is unclear how these companies measure that.

Are entry-level IT jobs at risk?

Stanford Digital Economy Lab researchers found employment for software developers aged 22-25 declined nearly 20% from its peak. Older age groups saw expansion.

Does AI make developers faster?

Not always. A METR trial found experienced open-source developers took 19% longer to complete issues with AI tools, though METR cautioned against generalizing that result.

Which IT jobs are projected to decline?

A summary of BLS 2025-35 projections says three computer occupations are projected to decline, including computer support specialists, as organizations use chatbots for troubleshooting. Software developers are projected to grow 10%.

Sources

Every number, quote and example on this page was checked by script against the sources below before publishing: quotes must match the source word for word, and figures must appear in a source or in BLS data. Our verdict is editorial judgment. Spotted an error? Tell us.

  1. U.S. Bureau of Labor Statistics: Occupational projections, 2025-35 (Table 1.2)
  2. AI Job Displacement Statistics 2026: Layoffs | Uvik Software (uvik.net)
  3. List of Companies Announcing AI-Driven Layoffs - Programs.com (programs.com)
  4. The Software Engineer's Survival Guide to the AI Era (2026) (bdtechjobs.com)
  5. Microsoft CEO says up to 30% of the company's code was written by AI | TechCrunch (techcrunch.com)
  6. How Much Code Is Written by AI? Statistics 2026 (eltexsoft.com)
  7. How Much Code Is Written by AI in 2026? The Real Numbers - Second Talent (secondtalent.com)
  8. Artificial intelligence, information technology, and employment, 2024–34 : The Economics Daily : U.S. Bureau of Labor Statistics (bls.gov)
  9. Will AI Replace Software Engineers? What the 2026 Data Shows | Extern (extern.com)
  10. AI and Tech Jobs: Which Roles Will Grow or Decline by 2035? (golabstech.com)
  11. Will AI Take My Job? What the 2026 US Data Shows (techjournal.org)
  12. AI impacts in BLS employment projections : The Economics Daily : U.S. Bureau of Labor Statistics (bls.gov)
  13. Computer Programmers (bls.gov)
  14. The government's new 10-year job outlook names AI in 48 occupations, up from 38 · Vocational Index (vocationalindex.com)
  15. Experienced developers were 19 percent slower with AI tools in a METR trial (scienceblog.com)
  16. Measuring the Impact of Early-2025 AI on Experienced Open-Source Developer Productivity (metr.org)
  17. METR randomized trial finds experienced developers 19% slower when using AI coding tools — One Year Ago in AI (oneyearago.ai)
  18. METR 2026: AI Made Developers 19% Slower. Is That Still True? (ingenire.com)
  19. Canaries in the Coal Mine? Six Facts about the Recent ... (digitaleconomy.stanford.edu)
  20. Canaries Dashboard - Stanford Digital Economy Lab (digitaleconomy.stanford.edu)
  21. Anthropic Economic Index: AI’s Impact on Software Development (tagteam.harvard.edu)
  22. Introducing the Anthropic Economic Index (anthropic.com)
  23. Ninth Circuit Rejects DMCA Copilot Theory | News | Haynes Boone (haynesboone.com)
  24. Nothing to Remove: Ninth Circuit Holds Copilot Creates New Code, Not Stripped Copies (fenwick.com)
  25. Ninth Circuit Limits DMCA Claims Against Generative AI: Key Legal Findings and Implications for IP Teams - Lexology (lexology.com)

The White-Hat Brief

Weekly: what the machines learned, and how to stay ahead

One email a week. Verified stories, plain-language defenses, no hype. Unsubscribe anytime.