AI boosted: seeing through new eyes
After my last post on ‘seeing through new eyes,’ someone asked me on Linkedin if I had used AI to write my personas, or if they were ‘hand crafted.’ In fact they were hand crafted, but it made me curious how AI would handle the same situation. So I used Gemini to develop three personas.
I told Gemini, ’I am running an ideation session and I want to create some personas to help me see the problem through their eyes. For each of three personas, I want to know what kind of knowledge they have and how they think about things. What might be some of the characteristics they have that color how they see things? what drives their behaviours? The three that I'd like to develop are: fashion designer, personal trainer, and homeless person.
I should note that I accidentally hit ‘return’ before typing the three categories for personas at the end. Left to its own devices, Gemini created three personas for:
tech savvy traditionalist
collaborative creator
data-driven analyst
I have to say these three all sounded similar (like they work next to each other in the same office) and too vague and generic to be of much use. They might serve as ‘user personas’ for some new app, but not for coming up with a range of really different ideas. I stand by my recommendation in the original post to pick more distinct categories.
So how did Gemini do compared to me?
Keep in mind that I did do some cursory internet searching to create my personas. In the tables below I’ve put my results and Gemini’s results side by side.
What I observe is that Gemini’s personas are longer, and more detailed, which could be helpful. (In my post I was of course also conscious of my reader’s time!) I was less impressed with how Gemini applied these results to create really different or unusual new ideas for ‘engagement in space exhibits.’
Although there were some interesting ideas, I found them all to be very literal, and sometimes using the persona as the ‘user’ or ‘client’ rather than as a way of seeing. For example from the personal trainer persona, I took the concept of ‘reaching goals’ to ask the broader question ‘how can these exhibits help visitors reach their goals’, whereas Gemini came up with purely fitness-oriented ideas.
In employing AI to do this kind of work, I might let AI create the personas that I have specified, but I would definitely do a hand crafting ‘first pass’ on creating new ideas from ‘seeing through their eyes’. I worry that going right to the AI’s ideas, might sway the results toward these more literal interpretations and miss the transformative ideas that come from taking more abstract concepts from one context to another.
Beyond that, I wrote before about how AI might be helpful in facilitation in this post. Then I also found that AI (in that case ChatGPT) gave kind of flat responses. This strikes me as a natural result of training these models on what has come before and responding to queries using the average of what has come before. I find in working with AI, you have to do a lot of iteration and direction to get useful material, but it definitely does have its purpose for some kinds of work.
If you’re looking for other ways to ‘see through new eyes,’ this column about different ‘thinking styles’ will also be useful.
What is your experience so far with AI in facilitation and innovation-related settings? Here is the side by side comparison of me and AI for an outline persona, and the ideas that emerged from ‘seeing through those eyes’:
