Software Developers

Software Developers score 87 out of 100, which is very high exposure to AI. That ranks #19 of 923 US occupations, where #1 is the most exposed. Of the 19 tasks in this job, AI could speed up 15 on its own and 3 more with extra software built on top of it. In April–May 2026, 11 of these tasks showed up in real conversations with Claude, Anthropic's AI assistant.

87/ 100

Very high

Rank: #19of 923 jobs
US workers: 1,687,890BLS, May 2025
Median pay: $135,980a year
In real AI use: 11 of 19tasks, April–May 2026

Free report

Follow how AI is changing software developers work.

Subscribe to the weekly report and get The AI Job Exposure Report (PDF) right away: the most and least exposed jobs, exposure by field, and the AI-resilient jobs that pay well.

Free · one email a week · unsubscribe anytime

Task by task

Each task this job involves, as listed by O*NET, rated on whether a large language model like GPT-4 could cut the time it takes by at least half at the same quality. Tasks marked Seen in real AI use appeared in conversations with Claude in April–May 2026 (Anthropic Economic Index).

AI can speed this up on its own: 15 (79%)
AI can speed this up with extra software: 3 (16%)
Little or no AI speed-up: 1 (5%)

Analyze information to determine, recommend, and plan installation of a new system or modification of an existing system.AI can speed this up on its own · Seen in real AI use
Analyze user needs and software requirements to determine feasibility of design within time and cost constraints.AI can speed this up on its own · Seen in real AI use
Confer with systems analysts, engineers, programmers and others to design systems and to obtain information on project limitations and capabilities, performance requirements and interfaces.AI can speed this up on its own · Seen in real AI use
Coordinate installation of software system.AI can speed this up on its own
Design, develop and modify software systems, using scientific analysis and mathematical models to predict and measure outcomes and consequences of design.AI can speed this up on its own · Seen in real AI use
Determine system performance standards.AI can speed this up on its own
Develop or direct software system testing or validation procedures, programming, or documentation.AI can speed this up on its own · Seen in real AI use
Modify existing software to correct errors, adapt it to new hardware, or upgrade interfaces and improve performance.AI can speed this up on its own · Seen in real AI use
Monitor functioning of equipment to ensure system operates in conformance with specifications.AI can speed this up on its own · Seen in real AI use
Obtain and evaluate information on factors such as reporting formats required, costs, or security needs to determine hardware configuration.AI can speed this up on its own · Seen in real AI use
Prepare reports or correspondence concerning project specifications, activities, or status.AI can speed this up on its own · Seen in real AI use
Specify power supply requirements and configuration.AI can speed this up on its own
Store, retrieve, and manipulate data for analysis of system capabilities and requirements.AI can speed this up on its own · Seen in real AI use
Supervise and assign work to programmers, designers, technologists, technicians, or other engineering or scientific personnel.AI can speed this up on its own
Supervise the work of programmers, technologists and technicians and other engineering and scientific personnel.AI can speed this up on its own
Confer with data processing or project managers to obtain information on limitations or capabilities for data processing projects.AI can speed this up with extra software
Consult with customers or other departments on project status, proposals, or technical issues, such as software system design or maintenance.AI can speed this up with extra software · Seen in real AI use
Recommend purchase of equipment to control dust, temperature, or humidity in area of system installation.AI can speed this up with extra software
Train users to use new or modified equipment.Little or no AI speed-up

What you can do about it

Get a personal plan for what to learn next.

Skill Half Life turns this score into a 90-day plan for your situation: your realistic options, the skills that carry over, and training that fits your time and budget. Coming soon.

See your options

Similar jobs, lower exposure

Related occupations that score at least 10 points lower.

Other jobs in this field

Computer and Mathematical occupations close to this one.

All computer and mathematical jobs and AI

Questions about software developers and AI

Will AI replace software developers?

Probably not entirely, but the job is likely to change a lot. Software Developers score 87 out of 100 (very high exposure), #19 of 923 US jobs. AI could speed up 18 of the job's 19 tasks, 15 of them on its own, while 1 see little or no speed-up. Exposure means the work is likely to change, not that the job will disappear.

Which software developers tasks can AI do?

The most exposed tasks include "Analyze information to determine, recommend, and plan installation of a new system or modification of an existing system"; "Analyze user needs and software requirements to determine feasibility of design within time and cost constraints"; "Confer with systems analysts, engineers, programmers and others to design systems and to obtain information on project limitations and capabilities, performance requirements and interfaces".

Which software developers tasks are hardest for AI?

Tasks with little or no AI speed-up include "Train users to use new or modified equipment".

Are people already using AI for this work?

Yes. In April–May 2026, 11 of the 19 tasks showed up in real conversations with Claude, according to the Anthropic Economic Index.

How many people work as software developers, and what do they earn?

About 1,687,890 people in the US, with median pay of $135,980 a year (BLS, May 2025).

How to read this score. Exposure measures how much of the work AI could speed up, not whether the job will disappear. A highly exposed job can change shape, with people spending less time on routine tasks and more on the rest. The score counts tasks AI can speed up on its own in full, and tasks that need extra software at half weight.

The ratings come from published research based on GPT-4 (2023). To show what's happening now, each task is also checked against the latest Anthropic Economic Index, which records which tasks people actually use AI for. That data is refreshed automatically when Anthropic publishes a new release. Read the full method.