Got into an argument at a Seattle meetup about AI writing code, then I looked up the actual numbers
Last Thursday I was at a small AI meetup in Seattle and this guy kept saying AI will replace most coders in 2 years. I pushed back and he told me to go look at the data, so I did. Turns out in a 2023 study from Stanford, devs using an AI assistant finished tasks about 55% faster, but they also shipped noticeably more bugs in the code. That surprised me because everyone just talks about the speed and never the quality part. I still think AI is huge, but acting like it's a straight swap for a whole job feels way off. Anyone here actually track bug rates on their team after adding AI tools, or is it just me noticing this?