AI / Thinking / Human agency
AI should push the frontier of our thinking
The most interesting use of AI is not getting an answer faster. It is reaching questions we could not hold clearly before.
Most conversations about AI begin with speed. A model can draft in seconds, search across more material, and generate ten directions while a person is still arranging the first. That is useful, but speed alone is a small ambition.
The more interesting possibility is that AI can extend the distance of a thought. It can help us compare ideas from fields that rarely meet, expose assumptions hidden inside a plan, and imagine consequences before we commit resources. Used this way, the model is less like an answer machine and more like a surface that makes our thinking visible.
That distinction matters. If we ask only for finished answers, our judgment slowly leaves the process. If we ask the system to challenge a premise, show competing explanations, or identify what evidence would change a conclusion, we stay responsible for the direction. The machine expands the search space; the person still decides what is worth pursuing.
Frontier thinking also requires disagreement. A useful system should not merely reflect the user’s first idea with more polished language. It should reveal the strongest objection, the missing variable, and the uncomfortable second-order effect. That friction can turn a vague conviction into a position we actually understand.
The opportunity is not to remove thinking from work. It is to make deeper thinking affordable in more moments: before a product decision, during research, while learning a difficult subject, or when an early idea feels promising but incomplete.
I want to build tools that reward that relationship. They should leave people with stronger questions, clearer reasoning, and more ownership of the final decision. The frontier moves when technology increases human agency rather than merely producing more output.