Announcing the ๐ฎ๐ฌ๐ฎ๐ฑ ๐ฆ๐๐ฎ๐๐ฒ ๐ผ๐ณ ๐๐ ๐๐ผ๐ฑ๐ถ๐ป๐ด, a study of AI software tool adoption, proficiency, and transformation among professional software developers globally. Key findings:
๐๐ ๐๐ผ๐ณ๐๐๐ฎ๐ฟ๐ฒ ๐ฎ๐ฑ๐ผ๐ฝ๐๐ถ๐ผ๐ป ๐ถ๐ ๐ป๐ผ๐ ๐ป๐ฒ๐ฎ๐ฟ๐น๐ ๐๐ป๐ถ๐๐ฒ๐ฟ๐๐ฎ๐น ๐ฎ๐บ๐ผ๐ป๐ด ๐ฑ๐ฒ๐๐ฒ๐น๐ผ๐ฝ๐ฒ๐ฟ๐: 98% of surveyed developers use AI coding tools several times a week. That being said, there is still some resistance among senior developers with them reporting less adoption.
๐๐ ๐ฐ๐ผ๐ฑ๐ถ๐ป๐ด ๐๐ผ๐ผ๐น๐ ๐ฎ๐ฟ๐ฒ ๐๐ถ๐ฑ๐ฒ๐น๐ ๐๐๐ฒ๐ฑ ๐ณ๐ผ๐ฟ ๐ฎ ๐๐ฎ๐ฟ๐ถ๐ฒ๐๐ ๐ผ๐ณ ๐๐ฎ๐๐ธ ๐๐๐ฝ๐ฒ๐: There is strong adoption with โexplain thisโ workflows, debugging, test generation, document generation, UI development, and code review. Security use cases however trail behind.
๐๐น๐ฎ๐๐ฑ๐ฒ ๐๐ผ๐ฑ๐ฒ ๐ฎ๐ป๐ฑ ๐๐๐ฟ๐๐ผ๐ฟ ๐ฑ๐ผ๐บ๐ถ๐ป๐ฎ๐๐ฒ ๐ฎ๐บ๐ผ๐ป๐ด ๐๐ผ๐ผ๐น ๐ฎ๐ฑ๐ผ๐ฝ๐๐ถ๐ผ๐ป ๐๐ถ๐๐ต ๐ฎ ๐น๐ผ๐ป๐ด ๐๐ฎ๐ถ๐น ๐ผ๐ณ ๐ผ๐๐ต๐ฒ๐ฟ ๐ฝ๐น๐ฎ๐๐ณ๐ผ๐ฟ๐บ ๐๐๐ฎ๐ด๐ฒ: These two have emerged as the coding platform category winners, though many developers are experimenting with other tools. Dedicated code review platforms still arenโt as widely adopted.
๐๐ถ๐ณ๐ณ๐ฒ๐ฟ๐ฒ๐ป๐ ๐ฐ๐ฎ๐๐ฒ๐ด๐ผ๐ฟ๐ถ๐ฒ๐ ๐ผ๐ณ ๐๐ผ๐ณ๐๐๐ฎ๐ฟ๐ฒ ๐ฒ๐ป๐ด๐ถ๐ป๐ฒ๐ฒ๐ฟ๐ถ๐ป๐ด ๐ฎ๐ฟ๐ฒ ๐๐ฒ๐ฒ๐ถ๐ป๐ด ๐๐ป๐ฒ๐๐ฒ๐ป ๐ฒ๐ณ๐ณ๐ถ๐ฐ๐ถ๐ฒ๐ป๐ฐ๐ ๐ด๐ฎ๐ถ๐ป๐: Frontend use cases are seeing the largest productivity gains. However in other software categories like infrastructure, mobile, and design the gains are not as widely perceived.
๐๐ฒ๐ป๐ฒ๐ฟ๐ฎ๐น ๐๐ฎ๐๐ถ๐๐ณ๐ฎ๐ฐ๐๐ถ๐ผ๐ป ๐๐ถ๐๐ต ๐๐ ๐ฐ๐ผ๐ฑ๐ถ๐ป๐ด ๐๐ผ๐ผ๐น ๐๐ฒ๐๐๐ฝ ๐ถ๐ ๐พ๐๐ถ๐๐ฒ ๐ต๐ถ๐ด๐ต: Developers are mostly happy with their AI coding tool setups though satisfaction levels drop at larger companies.
๐๐ ๐ฐ๐ผ๐ฑ๐ถ๐ป๐ด ๐๐ผ๐ผ๐น๐ ๐๐๐ถ๐น๐น ๐ต๐ฎ๐๐ฒ ๐ผ๐ฝ๐ฒ๐ป ๐ฝ๐ฟ๐ผ๐ฏ๐น๐ฒ๐บ๐ ๐๐ต๐ฎ๐ ๐ฑ๐ฒ๐๐ฒ๐น๐ผ๐ฝ๐ฒ๐ฟ๐ ๐๐ฎ๐ป๐ ๐๐ผ ๐๐ฒ๐ฒ ๐ถ๐บ๐ฝ๐ฟ๐ผ๐๐ฒ๐ฑ: Hallucination and lack of customization are reported as some of the biggest AI software failure modes. Of additional note, developers say context switching among tasks has become worse with AI coding.
Full link to report:
stateof.themodernsoftware.deโฆ