You're Probably Using AI Like an MBA
sachinkamdar.com
sachinkamdar.com
My take at work is that I'm not running much faster, but I am getting better quality. Some of it is my attitude, but with AI I am more likely to go back and forth and ask things until I really understand what is going on, write tests even when it is a hassle to write tests, ask the IDE questions about the dependencies I use so I can really understand how they work, try two or three possible solutions and pick the best, etc.
When it comes to things like that memory leak it is very hit and miss. If you give it try it might solve it, it might not. It's worth trying. But you can't count on something like that working all the time.
But you're describing exactly the shift that matters. You're not running faster, you're getting better quality. You're more likely to understand dependencies, write tests, try multiple solutions. That's the actual productivity gain.
The marshmallow challenge point isn't about whether AI makes you 10x faster. It's about the mindset shift. The MBAs didn't lose because they were slower. They lost because they spent their time planning the perfect approach instead of iterating.
The memory leak example from Boris Cherny isn't about AI being reliable. It's about his coworker not having the baggage of "this is how you debug memory leaks." They just tried asking Claude first. Sometimes it works, sometimes it doesn't. But the willingness to try it first is what creates the gap.
I don't think about AI tools a lot. I use Junie because it is integrated with my favorite IDE and like sending more money Jetbrain's way. I don't read blogs or tweets about AI coding.
What I do do is try little things that are always oriented to the work in front of me. I work for an MBA who is great at what he does but when he tried "vibe coding" he got nowhere and had that similar feeling of puzzlement from the gap between the results he got and the results that influencers say they are getting that a lot of people express. I've learned AI assisted coding by doing and from square one I realized it was going to work some of the time and fail some of the time and have always prioritized not getting stuck. It's certainly fair to make some wild (but well-formed) request and see what you can get.