Computer programmers

Create and modify code or scripts that simplify software development

Creating scripts and automation code is a sweet spot for current AI systems. These tasks typically involve well-defined inputs/outputs, standard patterns, and limited scope—exactly where LLM-based coding tools perform at or above median human level. AI can generate build scripts, deployment automation, and developer tooling with minimal human intervention beyond specification and validation.

BLS evidence: Programmers 'create, modify, and test code or scripts in software that simplifies development.'

Utilize code libraries to simplify writing and improve efficiency

AI coding assistants have comprehensive knowledge of popular code libraries and can suggest appropriate library functions, generate correct API calls, and compose library components effectively. This task is largely pattern-matching and documentation lookup, both areas where AI excels, requiring only light human verification of the suggested approach.

BLS evidence: Programmers use code libraries, which are collections of independent lines of code, to simplify their writing and improve their efficiency.

Write programs in various computer languages such as C++ and Java

AI coding assistants like GitHub Copilot, GPT-4, and Claude can now write substantial blocks of functional code in C++, Java, and other languages from natural language descriptions or partial code context. While human review and integration remain important, the core act of translating requirements into syntactically correct, logically sound code is increasingly automated, matching the anchor example of a developer writing a CRUD feature.

BLS evidence: Computer programmers write, modify, and test code and scripts that allow computer software and applications to function properly.

Update and expand existing programs

Expanding existing programs requires understanding legacy codebases and making coherent additions—tasks where AI code assistants now demonstrate strong capability through context-aware suggestions and refactoring. AI can analyze existing code structure, propose extensions, and implement features consistent with established patterns, though human oversight ensures architectural coherence.

BLS evidence: The duties section explicitly lists 'Update and expand existing programs' as a primary task.

Test programs for errors and fix faulty lines of computer code

AI systems excel at identifying syntax errors, common logic bugs, and code smells through static analysis and can suggest fixes with high accuracy. Modern AI can debug stack traces, identify root causes, and propose corrections. However, complex runtime bugs in production systems and subtle edge cases still benefit significantly from human expertise, placing this slightly below fully autonomous execution.

BLS evidence: Programmers run tests to ensure that newly created applications and software produce the expected results and check the code for mistakes.

Rewrite programs to work on different system platforms

Porting code across platforms involves understanding platform-specific APIs, dependencies, and constraints. AI can handle many standard conversions and identify platform-specific issues, but complex platform differences, performance optimization, and testing across environments still require substantial human involvement to ensure functional equivalence.

BLS evidence: Programmers typically need to rewrite their programs to work on different system platforms, such as Windows or OS X.

Design programs when duties overlap with software developers

Program design requires architectural thinking, understanding user needs, making trade-offs between competing concerns, and long-term maintainability judgments. AI can propose design patterns and generate architectural sketches, but the strategic decision-making and stakeholder alignment aspects mean humans remain central to the task, with AI providing substantial assistance.

BLS evidence: When overlap occurs with developers, programmers may be required to take on tasks typically assigned to developers, such as designing programs.

Learn new programming languages and technology upgrades through continuing education

While AI can provide tutorials, answer questions, and generate practice exercises for learning new technologies, the actual learning process—building mental models, developing intuition, and integrating knowledge—remains a human cognitive activity. AI accelerates learning but doesn't eliminate the need for programmers to personally acquire and internalize new skills.

BLS evidence: To keep up with changing technology, computer programmers may take continuing education classes and attend professional development seminars.

Collaborate with software developers and team members on large projects

Collaboration involves real-time communication, negotiation of technical approaches, understanding team dynamics, coordinating schedules, and building consensus—fundamentally human activities. While AI can assist with documentation or summarizing discussions, the interpersonal and coordination aspects of team collaboration remain heavily human-driven.

BLS evidence: Programmers work closely with software developers and must coordinate work on large projects with team members and managers.