Reasoning Models are amazing at Pattern Recognition
February 10, 2025
I don’t think we give enough credit to AI models—they’re remarkably skilled at uncovering hidden patterns that often go unnoticed.
Take ChatGPT o1, for example. Unlike ChatGPT 4o, it’s designed for complex tasks, where you can drop a big prompt and give detailed inputs. You can then ask, “Do you need anything else?” and it’ll pinpoint what's missing, asking for more specifics to refine its response.
The intelligence here is like magic. The interaction in a way feels completely different from earlier models. As compute increases, the hard problems for AI will become routine in 6 months or so.