Yejun Su

Be more ambitious in the AI age

Note of How to be more ambitious: A tutorial for humans with models.

Focus on outcomes 

Choosing the right outcome matters more than executing the task.

Using the Piano setup as an example, the equivalent progression would be: “fix this crash” → “investigate recurring crashes” → “keep released versions reliable.”

Go deeper 

Tackle the problem with the methods, tools, and rigor of a top professional—not merely produce a professional-sounding answer.

I want a deeper version of this task, not a bigger one.
Before rewriting it:
(1) say what a top professional in this field would do here
    that I'm not asking for;
(2) name the rawest material they would start from instead
    of what has already been digested for me;
(3) name the specific methods and tools they would use;
(4) say what, other than you or me, would show the result is right.
Then rewrite my prompt to ask for exactly that.

Draw across fields 

My goal is [outcome], with these constraints: [constraints].
What other fields could help solve this beyond the obvious one?
Choose up to three that would materially improve the result.
For each, explain its contribution in plain language
and name a concrete method we could apply.
Then use the useful methods—not just expert personas—and explain
how we can check whether they helped.

Build self-efficacy 

Build the belief that “I can figure out how to accomplish this” through attempting hard things, seeing others succeed, and allowing yourself to fail. It’s not that confidence guarantees success; it makes you more likely to attempt and persist at something you could succeed at.