Skip to content
EmergedDetailed profileemerged c.1950

Software Developer

Also: Programmer · Software engineer · Coder

Designs, writes, tests and maintains software.

1926

A · The occupation 100 years ago

Did not exist. Programming emerged in the late 1940s, initially done largely by women who had been human computers.

Historical imagery: not yet included. Images are added only after licensing (public domain or explicit permission) is confirmed.

1926–2026

B · The evolution

  1. 1948

    Programming is born

    Early stored-program computers required programming, much of it done by women.

  2. 1957

    High-level languages

    FORTRAN (1957) and COBOL (1959) made programming more productive.

  3. 1981

    The PC software industry

    Personal computers created a mass market for software.

  4. 1995

    The web era

    Web development became a mass occupation.

  5. 2023

    AI pair programmers

    In an experiment, developers with Copilot completed a task 55.8% faster.

What disappeared

  • No major tasks fully disappeared.

What technology took on

  • • Writing code (from c.2022)
  • • Debugging and testing (from c.2022)
  • • System design and architecture (from c.2024)
  • • Documentation (from c.2023)

What remained fundamentally human

  • • Requirements and stakeholder work
  • • Security and reliability responsibility

2026

C · The occupation today

One of the largest professional occupations. AI coding assistants are ubiquitous; controlled experiments show large speed-ups on defined tasks, and AI agents increasingly write and test code. Entry-level hiring has become more uncertain.

Required skills

  • Programming
  • System design
  • Problem solving
  • Communication

Typical tools & software

  • IDEs
  • AI coding assistants
  • Cloud platforms

Demonstrated today

Technology already used for these tasks (documented in 2015–2026)

  • • Writing code
  • • Debugging and testing
  • • System design and architecture
  • • Documentation

Speculative capability

Not yet demonstrated at scale — these appear only in scenarios

  • • Supervising AI coding agents (central c.2026)
  • • AI system governance (central c.2030)

Barriers to automation (editorial assessment, 0–1)

These factors delay or cap automation in the model. Strong underlying demand partly offsets very high exposure.

  • Regulation & licensing0.20
  • Physical environment0.05
  • Human accountability0.50
  • Social acceptance0.20
  • Economic viability0.80
  • Underlying demand trend+0.40

2030→2126

D · The future

Scenarios, not predictions. Confidence falls with the horizon.

2050 · CENTRAL SCENARIO

Most code generated by AI; developers specify, architect, review, secure and take responsibility for systems. Demand for software keeps growing.

Low confidence

2126 · CENTRAL SCENARIO

Software creation may be largely autonomous; human roles in defining goals, governance and novel systems.

Speculative

2126 · CONSERVATIVE

Developers remain central with very powerful tools.

2126 · ACCELERATED

Programming as a mass occupation largely gone; a small community oversees autonomous development.

0255075100192619501975200020262050207521002126NOWscenarios →← reconstructed
ConservativeCentralAcceleratedTask transformation index (0–100) · not a probability of job loss
Structured forecast per horizon and scenario (model task-curve-v1). Employment outlook is a qualitative direction, not a jobs number.
YearScenarioTransformationEmployment outlookSkills outlookConfidence
2030Conservative23Growing
Central22GrowingCore human skills remain in requirements and stakeholder work, security and reliability responsibility; new capabilities needed for supervising ai coding agents, ai system governance; plus fluency directing and auditing AI tools.Moderate
Accelerated30Stable, transforming
2050Conservative22Growing
Central37Stable, transformingCore human skills remain in requirements and stakeholder work, security and reliability responsibility; new capabilities needed for supervising ai coding agents, ai system governance; plus fluency directing and auditing autonomous systems.Low
Accelerated54Stable, transforming
2075Conservative37Stable, transforming
Central57Stable, transformingCore human skills remain in requirements and stakeholder work, security and reliability responsibility; new capabilities needed for supervising ai coding agents, ai system governance; plus fluency directing and auditing autonomous systems.Speculative
Accelerated66Contracting
2126Conservative52Stable, transforming
Central64ContractingCore human skills remain in security and reliability responsibility; new capabilities needed for supervising ai coding agents, ai system governance; plus fluency directing and auditing autonomous systems.Speculative
Accelerated70Contracting

Salary implications: not modelled. No credible occupation-level wage forecasts exist for these horizons, so we do not publish one.

Tasks

E · Occupational transformation

  • Routine cognitive
  • Non-routine analytic
  • Creative production
  • Interpersonal & care
  • Judgement & accountability
2050
37/100Model output · Central scenarioLow confidence
  • Human-led · 2
  • Assisted · 3
  • Automated · 1
  • New task · 2
  • Retired · 0
Tasks in 2050
TaskMachine involvement
Writing codeAssisted
Debugging and testingAssisted
System design and architectureAssisted
Requirements and stakeholder workHuman-led
Security and reliability responsibilityHuman-led
DocumentationAutomated
Supervising AI coding agentsnewAssisted
AI system governancenewHuman-led

How this is derived: tasks are classified with the routine/non-routine framework (Autor, Levy & Murnane 2003); each task type follows a capability curve per scenario, delayed by this occupation's adoption barriers (regulation, physical environment, social acceptance, economics) and capped by accountability. The index is a share of task content, not the probability that the job disappears. Full methodology →

Evidence

5 sources cited

The Impact of AI on Developer Productivity: Evidence from GitHub CopilotSida Peng; Eirini Kalliamvakou; Peter Cihon; Mert Demirer · arXiv:2302.06590 · 2023-02In a controlled experiment, developers with GitHub Copilot completed a JavaScript HTTP-server task 55.8% faster than the control group.Caveat: Single, well-specified task; authors affiliated with GitHub/Microsoft.Working paperID peng-2023Original ↗RegisterO*NET OnLineUS Department of Labor, Employment and Training Administration · continuously updatedDetailed descriptors for roughly 900+ US occupations: tasks, tools and technology, skills, work context.Statistical datasetID onetOriginal ↗RegisterOccupational Outlook HandbookUS Bureau of Labor Statistics · continuously updatedOfficial US descriptions of what workers do, pay, and ten-year employment projections by occupation.Government reportID bls-oohOriginal ↗RegisterGPTs are GPTs: An Early Look at the Labor Market Impact Potential of Large Language ModelsTyna Eloundou; Sam Manning; Pamela Mishkin; Daniel Rock · arXiv:2303.10130 (later published in Science, 2024) · 2023-03Estimates around 80% of the US workforce could have at least 10% of their tasks affected by LLMs, and about 19% at least half of their tasks; exposure is higher for higher-wage occupations.Caveat: Exposure measures where LLMs could reduce task time, not whether jobs are lost. Authors include OpenAI staff.Working paperID eloundou-2023Original ↗RegisterO*NET OnLine — Software Developer (15-1252.00)U.S. Department of Labor, Employment and Training Administration (O*NET) · Continuously updatedOccupation-specific tasks, tools and technology, skills and work context used to structure this profile's 2026 task list.Caveat: US-specific occupational definitions.ClassificationID onet-15-1252.00Original ↗Register

Profile last updated 2026-11-17.