User Research

Anyone seen great 2024 case studies on how teams are using AI to generate personas from interview transcripts?

I would love to be able to
1. Upload interview transcripts (w/ highlights and tags if necessary)
2. Accept/reject AI-proposed persona dimensions
3. Accept/reject AI-proposed persona details (goals, challenges, expectations, delighters, etc.)

And at each accept/reject step, I'd love to see quotes the AI selected as evidence to generate that proposal, plus a list of quotes ranked by their relevance to the dimension/detail in question, but were ultimately ignored. The dream would be a nice UI that lets me make adjustments along the way and not lose track.

Anyone done something similar to this? I can't get very far with Claude/GPT today

research /openux || learning /manansh || writing manansh.xyz

Interesting, I wanna hear more! Let's talk about this next time we chat 😎

Haven’t seen any case studies!!!

The problem is that LLMs don’t really reason.

They do prediction.

So extracting novel, different insights from a dataset and turning them into robust personas isn’t what its designed to do.

I think that might change once we create the first AI system that does its own AI research, because then, it’s really quite trivial to get it to do user research.