Will AI replace Computer Programmers?

Many of your tasks are within reach of AI. The role will change, so it pays to act now.

Task exposure to today's AI

63/100

High exposure

This is the occupation's typical task mix. Your own score can be up to 30 points higher or lower depending on how you work: routine share, people contact, judgment and how much you already use AI.

Get your personal score

Will AI replace Computer Programmers?

Not entirely. Today's AI tools can take on about 63% of the typical work in this occupation, mostly: write, update, and maintain computer programs or software packages to handle specific jobs such as tracking inventory, storing or retrieving data, or controlling other equipment. The parts that stay human include: Consult with managerial, engineering, and technical personnel to clarify program intent, identify problems, and suggest changes.; Write, analyze, review, and rewrite programs, using workflow chart and diagram, and applying knowledge of computer capabilities, subject matter, and symbolic logic.; Perform or direct revision, repair, or expansion of existing programs to increase operating efficiency or adapt to new requirements..

People who learn to use AI for the automatable parts keep the edge. The 3-phase plan below shows where to start.

Which tasks AI can take on

The largest tasks in this occupation, their share of the working week, and how much of each today's AI tools can do.

TaskAutomatable

Write, update, and maintain computer programs or software packages to handle specific jobs such as tracking inventory, storing or retrieving data, or controlling other equipment.AI already used 78%

Write, analyze, review, and rewrite programs, using workflow chart and diagram, and applying knowledge of computer capabilities, subject matter, and symbolic logic.AI already used 75%

Correct errors by making appropriate changes and rechecking the program to ensure that the desired results are produced.AI already used 72%

Conduct trial runs of programs and software applications to be sure they will produce the desired information and that the instructions are correct. 70%

Perform or direct revision, repair, or expansion of existing programs to increase operating efficiency or adapt to new requirements.AI already used 60%

Consult with managerial, engineering, and technical personnel to clarify program intent, identify problems, and suggest changes. 17%

+ 11 more tasks (25% of the week) in the full test.

Your human edge

Consult with managerial, engineering, and technical personnel to clarify program intent, identify problems, and suggest changes.
Write, analyze, review, and rewrite programs, using workflow chart and diagram, and applying knowledge of computer capabilities, subject matter, and symbolic logic.
Perform or direct revision, repair, or expansion of existing programs to increase operating efficiency or adapt to new requirements.

AI tools to try

Coding assistants (GitHub Copilot, Claude Code, Cursor)ChatGPT / Claude for design and debuggingAI test and documentation generators

Related roles

Software Developers
Web and Digital Interface Designers
Computer Systems Engineers/Architects

A 3-phase AI learning plan

Phase 1

Pair programming with AI

Use a coding assistant for boilerplate, tests and refactoring.
Give it context: project rules, examples, small steps.
Review AI code like a colleague's pull request.

Project

Pick a legacy module you own, write a project rules file for a coding assistant, then have it add unit tests to reach 80% coverage and refactor one function - log hours vs your manual estimate.

Phase 2

AI in your systems

Call an LLM API with structured output.
Retrieval over your own documents or data.
Evaluate quality, cost and latency before shipping.

Project

Build an internal tool that answers questions over your team's own docs via retrieval and an LLM API returning structured JSON, and benchmark accuracy, cost per query and latency on 30 test questions.

Phase 3

Production-grade AI

Evals and regression tests for prompts.
Security: prompt injection, data leaks, permissions.
Monitoring and cost control.

Project

Add a 50-case eval suite, prompt-injection tests and token-cost dashboard to that tool, then report regression pass rate and cost per 1,000 calls before and after tuning.

Take the free 11-question test