Developers rely on AI tools but remain wary of trusting their output

Stack Overflow's survey collected 30,903 responses from developers in 169 countries. Among daily AI users, nearly a third spend four or more hours per workday with these tools, and about 80 percent use them at least an hour daily. Only 7 percent trust AI output for important work decisions, while 79 percent say source attribution is important for trusting technical answers.

EXPANDED:

Stack Overflow's poll gathered 30,903 replies from 169 nations. Of 10,926 daily AI users answering a time question, roughly four in five used such tools at least an hour per workday, and close to one-third spent four hours or more. Coding assistants or agents were reported by 66%, general chat tools by 63%, and automated agent workflows by 26%; 17% used no AI.

Trust remained conditional: only 7% relied on AI output for significant work choices, 48% trusted it when verification was easy, and 79% valued source attribution. The company's paid OverflowAPI requires attribution, while its OpenAI deal, which trained models on 15 years of user contributions, sparked backlash. Also, 52% learned a new coding skill or language this year, with AI code generators among common resources after technical documentation.

Count? First paragraph: Stack(1) Overflow's(2) poll(3) gathered(4) 30,903(5) replies(6) from(7) 169(8) nations(9). Of(10) 10,926(11) daily(12) AI(13) users(14) answering(15) a(16) time(17) question(18), roughly(19) four(20) in(21) five(22) used(23) such(24) tools(25) at(26) least(27) an(28) hour(29) per(30) workday(31), and(32) close(33) to(34) one-third(35) spent(36) four(37) hours(38) or(39) more(40). Coding(41) assistants(42) or(43) agents(44) were(45) reported(46) by(47) 66%(48), general(49) chat(50) tools(51) by(52) 63%(53), and(54) automated(55) agent(56) workflows(57) by(58) 26%(59); 17%(60) used(61) no(62) AI(63). Good.

Second: Trust(1) remained(2) conditi