
Reblogged by jsonstein@masto.deoan.org ("Jeff Sonstein"):
ldwg@mastodon.world ("Ldwg") wrote:
🔥 I enjoyed the read!
🎯 Key insight for me: We still need human judgment to spot truly valuable solutions.
🤷♂️ Similar pattern with OpenAI O3: Impressive for problems where it's easy to tell if the solution is correct.
🤔 My experience with letting GPT-4 and Claude improve a SQL migration: Both only added boilerplate.
✨ What's working nicely for me:
First chat: Explore approaches with AI (no code)
Fresh chat: Build out the best idea (code)