Software Developer: AI job impact
AI can help draft code, tests and documentation. Choosing requirements, integrating changes and accepting production risk still require engineering judgment.
AI applicability
0.278 / 1
How applicable AI was to this occupation’s work in Microsoft’s study. This is an index, not a probability.
Observed activity coverage
75.0%
Share of work activities above the study’s usage threshold, weighted by importance. Not the share of workers using AI.
Covers software development across applications and systems. Specialties and seniority can change the task mix.
Microsoft v1.1 · U.S. Bing Copilot conversations, Jan–Sep 2024. These indicators do not predict job losses and do not change with your task selections. Measurement details
Your daily tasks
Set how often you do each task to update your task mix, plan and tool suggestions.
Frequency weights: occasionally 1, regularly 2, often 3. These shares do not represent working hours, time saved or job-loss risk.
AI roles are AI IDE List editorial assessments of O*NET tasks, not measured automation rates. Partial automation requires review.
Implement routine code
Generate a small implementation from an explicit interface; inspect dependencies and run it in a test environment.
Task source
O*NET original task statement (English)
Design, develop and modify software systems, using scientific analysis and mathematical models to predict and measure outcomes and consequences of design.
O*NET 31.0, task 21667 · Importance 3.56 / 5. View occupation tasks
Draft unit tests
Generate boundary and failure cases from requirements; confirm the assertions catch a deliberately broken implementation.
Task source
O*NET original task statement (English)
Develop or direct software system testing or validation procedures, programming, or documentation.
O*NET 31.0, task 21669 · Importance 3.90 / 5. View occupation tasks
Write technical documentation
Draft documentation from reviewed code; run every example and verify the described behavior.
Task source
O*NET original task statement (English)
Develop or direct software system testing or validation procedures, programming, or documentation.
O*NET 31.0, task 21669 · Importance 3.90 / 5. View occupation tasks
Investigate system changes
Compare possible causes and migration steps; verify each assumption against logs and the existing system.
Task source
O*NET original task statement (English)
Analyze information to determine, recommend, and plan installation of a new system or modification of an existing system.
O*NET 31.0, task 21661 · Importance 3.61 / 5. View occupation tasks
Define requirements
Use AI to list ambiguous requirements; resolve tradeoffs with the people who own the product.
Task source
O*NET original task statement (English)
Analyze user needs and software requirements to determine feasibility of design within time and cost constraints.
O*NET 31.0, task 21662 · Importance 4.08 / 5. View occupation tasks
Design system interfaces
Explore architecture options; engineers decide reliability, security and operational tradeoffs.
Task source
O*NET original task statement (English)
Confer with systems analysts, engineers, programmers and others to design systems and to obtain information on project limitations and capabilities, performance requirements and interfaces.
O*NET 31.0, task 21664 · Importance 3.75 / 5. View occupation tasks
Set performance standards
Draft a benchmark plan; choose realistic workloads and acceptable latency with service owners.
Task source
O*NET original task statement (English)
Determine system performance standards.
O*NET 31.0, task 21668 · Importance 3.53 / 5. View occupation tasks
Discuss delivery and technical issues
Prepare a concise status draft; confirm commitments and explain unresolved risks directly to stakeholders.
Task source
O*NET original task statement (English)
Consult with customers or other departments on project status, proposals, or technical issues, such as software system design or maintenance.
O*NET 31.0, task 21665 · Importance 3.37 / 5. View occupation tasks
Your selections are saved in this browser. Sharing a link includes only the occupation and selected frequencies.
Your AI adaptation plan
Start with one recurring task. Test the output before expanding the workflow.
Review generated code
For: Implement routine code
- Prepare
- Start with a small task, a stated interface and acceptance criteria.
- Try
- Generate a small implementation from an explicit interface; inspect dependencies and run it in a test environment.
- Verify
- Inspect the diff, run independent tests and check dependencies before using the change.
Check whether tests detect failures
For: Draft unit tests
- Prepare
- Use a requirement and a known failing example, without showing the proposed implementation first.
- Try
- Generate boundary and failure cases from requirements; confirm the assertions catch a deliberately broken implementation.
- Verify
- Confirm the tests fail on the broken example and pass on the corrected behavior.
Verify AI-written documentation
For: Write technical documentation
- Prepare
- Supply approved source material, the intended reader and the required output format.
- Try
- Draft documentation from reviewed code; run every example and verify the described behavior.
- Verify
- Trace factual claims to the source and run any procedural examples.
Review AI-generated changes
Take a small issue, write acceptance tests before prompting, then review the diff and run the tests. Record which errors the tests missed.
Skills and career outlook
Skills in the occupation data
Importance · 1–5- Programming4.00 / 5
- Critical Thinking3.88 / 5
- Judgment and Decision Making3.62 / 5
- Reading Comprehension3.50 / 5
- Active Learning3.50 / 5
Top O*NET essential and transferable skills by importance. Ratings describe occupational requirements, not future demand.
U.S. employment projection
+10.2%
Projected employment change, 2025–2035
Software developers
BLS projects total employment change from many factors. This is not an AI-specific forecast or a UK projection.
BLS employment projectionsWhat to keep evaluating
Code that passes a small example can still break contracts, expose data or fail under load. Review changes against the system, not only the generated diff.
Tools to try with your tasks
Matched to the tasks you selected. Test on a small example and compare output quality and review effort.
Claude Code
ToolExplore a codebase and propose a bounded change with tests.
Try for: Implement routine code; Draft unit tests.
Claude Code commandsClaude Opus 5.5
ModelA model to evaluate on a small, verified example of this task. Compare quality and review effort before adopting it.
Try for: Implement routine code; Draft unit tests.
GitHub Copilot
ToolDraft code, test cases and documentation alongside the code you are reviewing.
Try for: Implement routine code; Draft unit tests.
GPT-6 Astra
ModelA model to evaluate on a small, verified example of this task. Compare quality and review effort before adopting it.
Try for: Implement routine code; Draft unit tests.
Sources and limits
Matched occupation: Software Developers (15-1252.00). Published research group: Software Developers (SOC 15-1252).
- O*NET 31.0 · CC BY 4.0
- Working with AI, v1.1 · CC BY 4.0
- BLS Employment Projections, 2025–2035 · U.S. government data
Microsoft’s study uses O*NET 29.0; the task descriptions here use O*NET 31.0. The original study score is preserved. Editorial suggestions are not results reported by these sources.
Read the methodology, attribution and update record