A research-backed study surveying 1,300+ developers and interviewing 22 AI-fluent practitioners explores how AI is reshaping the developer role. Developers' openness to AI varies by task type: they welcome AI for coding and ops work but want to retain control over relational and stakeholder work. AI fluency follows a four-stage journey from Skeptic to Strategist, with advanced users orchestrating multiple agents and front-loading context. Key tensions include a 'learning paradox' around prompting vs. engineering mastery, deskilling risks for junior developers, and accountability gaps as velocity increases. Three possible futures are outlined: AI as accelerator, orchestration-first engineering, or a dystopian 'Clerical Coder' model where humans rubber-stamp AI output at scale.
Table of contents
What developers actually want from AIHow AI fluency actually developsThe tensions that don’t have easy answersThree futures, and the choices that decide which we getSort: