Question

The biggest problem with AI coding assistants is that they create a lot of “buggy code”.

So, what’s the solution?

Make AI learn to fix its own mess

Alibaba built SWE-smith: an automatic bug generator + fixer that:

• Creates real bugs in GitHub repos

• Verifies them via unit tests

• Has AI agents fix them step-by-step

• Produces before/after examples at scale

Basically, don’t just have AI models generate code, but also fine-tune them to fix the bugs.

What do you think? Would this solve the problem?

Comments

It will help but it will take years to make it better. We are forgetting that AI is a language model, just a good pattern matching. It still doesn't think. It looks like it's thinking. And we should be grateful for that.

---

Very rightly said!

---

i think this would solve the problem.

---

To make more useful with current capacity what it have, we need to break things into small modueles then it handles very well.