Line of inquiry
Inquiring lines›How can we ensure AI systems remai…›How should autonomy and oversight…›this line of inquiry
Does AI coding assistance improve productivity and code quality across experience levels?
A broader line of inquiry — a family of 49 specific questions the research asks around this. Follow one into its inquiring-line page, or move sideways to a related line below.
Questions in this line of inquiry 49
Specific inquiring lines the field asks around this — ordered from the most general framing down to the most specific angle.
- Why do experienced developers benefit more from AI coding assistance?
- Do AI coding tools improve code quality alongside task speed?
- Does AI-assisted coding actually speed up experienced developers?
- Does AI coding assistance help junior developers close skill gaps?
- Does AI help more on small greenfield projects than mature codebases?
- Why do developer self-reports of AI speedups tend to be unreliable?
- What code quality issues persist even when AI agents produce working code?
- Does high-level design work benefit differently from AI than routine coding tasks?
- Why does AI code generation lag behind pattern-matching benchmarks?
- Does minimal code engagement during vibe coding harm students' long-term programming comprehension?
- How should AI agents time interruptions during active programming tasks?
- Do AI tools erode users' ability to oversee and understand systems they built?
- How should code authorship be measured in human-AI collaborative development?
- When do developers use heuristics versus detailed oversight in practice?
- What self-regulation practices do junior developers use when deciding to accept AI output?
- Why do experienced developers report slower task completion with AI assistance?
- Why did programmer headcount not shrink after AI coding tools arrived?
- Does prior IDE tool use predict stickiness with new coding assistants?
- What role should code review play in junior developer learning with AI?
- Why do novice engineers lose confidence in coding after using AI tools?
- How does staged code generation compare to inline autocompletion for learning outcomes?
- Does hint-based scaffolding reduce the temptation to skip code validation?
- When does complete code generation work better than partial solutions?
- Why do low-adoption countries use AI primarily for coding tasks?
- Why did early-2025 AI tools slow down experienced open-source developers?
- Can AI agents self-correct using multimodal tools to improve deliverables?
- How much time do developers spend verifying AI-generated code?
- Why do novices accept AI output without validation in vibe coding workflows?
- Why does evaluation of novel primitives require waiting for future reuse to show value?
- Can AI design tools like Figma Make show speedups when coding tools show slowdowns?
- When do students feel authentic ownership of code they co-created with AI?
- Why haven't AI agents replaced human code review workflows?
- How does validating code differ from writing code as a learning mechanism?
- What makes intermediate primitives matter more than final code execution success?
- What role did API access and coding tools play in the output surge?
- Can explanations alongside partial code prevent cognitive overload for learners?
- How do non-experts evaluate AI-generated outputs when they lack implementation expertise?
- How much time do developers spend reviewing and fixing AI code?
- How does vibe coding sit within broader human-in-the-loop development workflows?
- Why do AI-inserted text and code suggestions survive at different rates?
- What distinguishes improvised spreadsheet workflows from institutionalized enterprise software?
- What debugging behaviors signal that a user has abandoned the coding loop?
- How does skipping architectural decisions early create hidden technical debt in AI systems?
- Why does embedding research tools in coding assistants improve reliability?
- What makes consistency across code, tables, figures, and prose the hard part?
- How does prior coding experience change the way students use vibe coding tools?
- How much human time remains after an agent completes a test-passing pull request?
- How do skill libraries from human resources compare to hand-written skill libraries?
- How much build time does declarative configuration save versus custom code?