By ICG member John Habershon
My favourite local restaurant has a mains menu with just five choices — one vegetarian dish, one fish, and three meat dishes. I get a delicious mix of emotions when I pick up the menu — pleasure, reassurance and, yes, a little smugness. I know whatever I choose will be good. I can relax, ready to enjoy the meal, perhaps spending more time on the wine list.
Many studies of buying behaviour have concluded that too much choice creates confusion – even buyer paralysis.
I have now settled on just two AI tools to fulfil all my data analysis and presentation needs.
So how are you to choose your AI partners? There are countless AI tools to choose from, each having very specific functions, some claiming to do it all.
I started with ChatGPT and moved on to Claude for its default high data security and higher word limit, among other factors. Since then I’ve used it on several projects — notably a qualitative project with 30 IDIs and a quantitative project with 5.5K respondents. I have also gained views from users of a range of AI tools from participants in my AI workshops.
The days of the standalone empirical research projects are numbered, as is the notion of separate ‘desk research’. Preparation for conducting research will involve placing it in a wider canvas. Claude 3.7 Sonnet is very good at finding published studies on a topic. The latest Google NotebookLM can operate like a super Google. It lets you search the web directly from inside your notebook. When a user describes their topic, NotebookLM scans through hundreds of web sources in seconds. It picks out up to 10 of the most relevant ones, complete with annotated summaries that explain why each source might matter to your topic. With just a single click, you can pull them into your notebook and add them as part of your sources.
How you proceed with your analysis is left to you. Unlike the ‘comprehensive’ front-end qualitative research packages there is no set way to proceed, leaving the choices to your judgement and creativity (although Claude 3.7 Sonnet will offer suggestions).
Similarly the project will no longer end when results are offered in isolation. The tools will place your findings in the context of other studies and existing knowledge.
There is no longer a barrier between qualitative and quantitative analysis. Claude makes it easy to move on to a quantitative follow-up, from designing a questionnaire to creating a synthetic dataset with which to pilot it.
Finally, the communication of findings is breaking free from the tired Powerpoint deck. Interactive visuals of data can be shared, findings offered in a fun quiz format, or relayed as a audio podcast.
YouTube Videos
A Powerful Combination of Two AI Tools for Qualitative Research
Claude 3 7 Sonnet for Quantitative Analysis and Output