Article by ICG member Karen Cooper, Living Brands.
New year’s resolution anyone?
No, not the personal eat-less, exercise-more types, but the work ones…. like less printing, no slack after 7pm, and perhaps embrace AI? If the last one’s on your list, I did a little comparison of co-pilots for qual analysis that may help you on your way.
I’m no expert (don’t ask me to explain github or python), but I had a feeling that AI co-pilots could help me with labour intensive qualitative analysis.
My test compared human brain-only analysis, to brain plus three different co-pilots fed recordings of a recent qual project. To spare blushes I’m only going to reveal my favourite pilot, but for context one was on a well-established meeting platform, one on a specialised qual/community research platform and the third on a transcription platform.
Brain-only worked well (phew), but that’s because it was a small (4 focus group) UK only project and I could retain most of it in my head. Brain also caught the non-language cues and knew the cultural context intuitively and added them into analysis. So, brain’s not being replaced right now (phew again). However, I found one assistant helped me find deeper insights whilst also saving me time.
Disclaimer – I didn’t replicate prompts word-for-word, so perhaps not a completely scientific test, though I asked similar big and general questions first and followed up with specific questions digging deeper into a particular line of enquiry.
Two were disappointingly bland. Summaries and big themes were solid and quotes easily found to back each point up (a fabulous time saver compared to the old days of manually finding quotes in transcripts – yay!), but I got no real depth.
The magic happened on the transcription platform – name check: otter ai, when using its chat function. Perhaps because they have been in the language game for longest, it seemed to understand more obscure questions and became my trusted friend – I kept going back to ask further questions which genuinely helped my thinking.
My only gripe was the inability to use chat function across the whole project (can only do it by transcription session/group by group). So, otter, next iteration please add this function.
I know the market’s evolving like crazy and am sure there a many other assistants out there (check their privacy policies and weigh up different pricing), but I will be using Otter.aichat to help my thinking going forwards.
That’s one news year’s resolution I will keep.