> While we're at it, why not have the AI automatically detect all bugs and fix them? /s Why not? Nothing's going to catch all bugs, but there's plenty of low-

> While we're at it, why not have the AI automatically detect all bugs and fix them? /s Why not? Nothing's going to catch all bugs, but there's plenty of low-hanging fruit that could benefit from automated scanning. I think GitHub already does have LLM-suggested fixes for security issues[0], even. Same with merge conflicts - majority are not that complex/deep. Offering a sensible resolution seems within the reach of current models (with good integration and fine-tuning). Could maybe even have a model check and warn of the cases that ORT would currently merge uncritically despite producing broken code. [0]: https://github.blog/wp-content/uploads/2024/02/303457375-1a8933c2-9171-4eab-b08f-fc19fd5a3573.mp4#t=0.001?_=1 https://github.blog/wp-content/uploads/2024/02/303457375-1a8...