I had a pretty incredible vibe coding experience with o3 today. As I’m sure many of you have also had recently – whether with o3, or Claude or Gemini.
I was iterating on a problem with it over a couple of hours. I asked it to come up with an idea for a novel algorithm application for an advanced ML project. Then asking it to implement it. And write unit tests. I did static code analysis on the code it generated with the assistance of a separate instance of o3 before running anything. The I iterated over the unit tests it wrote – executing them, feeding back errors, and offering high level guidance.
This algorithm was sufficiently advanced that I am not entirely sure it is correct, even after this iteration cycle. I’ll probably need to do continued testing and analysis for a couple of days.
To put that another way: a computer generated a program in two hours that will take me two days to fully test and understand.
With sufficiently accurate predictions from a chatbot, it becomes less logical (or even possible!) for me to fully check all of its outputs – I just have to “vibe” it.
I think it’s possible that we humans might stop being able to reasonably verify the outputs of these computer systems. That’ll be an interesting moment for each of us.