1, #66 - The value of a well-stated problem
On questioning AI — Estimated Read Time: 52 seconds.
“A problem well-stated is a problem half-solved.”
This quote from inventor Charles Kettering has been on my mind a lot lately, especially in relation to ChatGPT.
Whether or not this much-hyped AI tool responds with a useful answer, I often find value in the process, because of the need to find the right question.
I heard someone say that these types of tools look like a conversation, but actually work like an equation: The key is figuring out what variables you need to insert to get the solution you want.
For me, this means that I often switch over to my ChatGPT tab and write and rewrite and edit and re-edit the prompt, adding as much specificity and context as I can. By the time I’m finished, I sometimes barely even read the response. Not because of the often-generic and sometimes-hallucinated responses—but because, by the time I’ve clearly laid out the problem, I can already get a sense for the solution.
So, although I avidly encourage all my friends to start practicing with AI, you might get a similar benefit from a far less technologically advanced method:
Trying to figure out a tricky problem? Try just writing down the challenge with as much detail as possible. You may be surprised what your non-artificial intelligence is able to develop, from there.
Insight from: Charles Kettering, by way of Matt Lerner’s (excellent) new book, Growth Levers.
I speak so much about being "problem" obsessed vs. "solution" obsessed as a product manager. I've never thought about how ChatGPT could be a useful teacher for this. I'm totally going to try this with my students this spring. Great post!
Love it!
There are two frameworks I use a lot for that:
SCQA: Situation, Complication, Question - Answer
TOSCA: Trouble, Owner, Success Criteria, Constraints and Actors
The first is leaner; the latter often shows that problems of specific people differ from the general problem (think of increasing revenue as a problem for a company - what is the problem of the CMO or the marketer) and that much information we need for a well-stated problem are the constraints and how we know it's solved.
I've stolen it from the book Cracked It.