Regarding "Don't rely on AI note-takers" comment, I agree that ChatGPT's summarization of transacriptions is currently poor.
However, this does not apply to ALL current AI tools. Specficially, there are User Research-specific tools that leverage exact trnscripts and provide links to source quotes and/or video clips in their summaries. I've briefly used Quallie.ai and Hey Marvin that do this and found them to be quite reliable. I know other tools do this too.
My point is - researchers should not avoid AI, but instead should be skeptical and try the tools themselves and check their sources.
Importantly, as Caitlin Sullivan shows in her AI Customer Research Analysis course, even ChatGPT o3, Gemini Advanced 2.5 Pro (experimental) or Claude Sonnet 3.7 reasoning models can - and should (!) - be clearly instructed, shown examples of what good looks like, prompted in chain of thought steps, and pushed to come up with a thorough, clearly structured analysis. Where each of the above models needs to be pushed differs.
I'm just in week 2 of her 4-week course. There is a ton to learn from her very extensive experience with specific AI tools for UX research as well.
Regarding "Don't rely on AI note-takers" comment, I agree that ChatGPT's summarization of transacriptions is currently poor.
However, this does not apply to ALL current AI tools. Specficially, there are User Research-specific tools that leverage exact trnscripts and provide links to source quotes and/or video clips in their summaries. I've briefly used Quallie.ai and Hey Marvin that do this and found them to be quite reliable. I know other tools do this too.
My point is - researchers should not avoid AI, but instead should be skeptical and try the tools themselves and check their sources.
Importantly, as Caitlin Sullivan shows in her AI Customer Research Analysis course, even ChatGPT o3, Gemini Advanced 2.5 Pro (experimental) or Claude Sonnet 3.7 reasoning models can - and should (!) - be clearly instructed, shown examples of what good looks like, prompted in chain of thought steps, and pushed to come up with a thorough, clearly structured analysis. Where each of the above models needs to be pushed differs.
I'm just in week 2 of her 4-week course. There is a ton to learn from her very extensive experience with specific AI tools for UX research as well.
Great point Pete - thank you for sharing. I 100% agree and have added those methods to my AI practice.
I want to give a shoutout to Tomer Sharon’s “rainbow spreadsheet.” It’s not only a practical note-taking technique but also a cross-team friendly one.