🔭 I wanted to see how well AI could help me generate ideas by borrowing UX/UI concepts/features from other industries.

Amazing. I personally believe some of my best ideas come from looking at what's being done in other industries. So, I was THRILLED when the results were super insightful, easy to build upon, etc.

Some of the ideas weren't as original, would be interesting to suggest more obscure industry comparisons.

https://docs.google.com/document/d/1-qIw0oMfCRie9m_zC1C0TU2FG46fPScULI58Lsp73bI/edit?usp=sharing

#aiexperimentlog #1linkedin #highsuccess

🔭 Restest of @amaya_arie6 "quick idea pitches + worst-case scenario teardown."

I used the NN/g GPT for this retest and I don't think it was as effective at creating those punchy, one-liners that Amaya generated. It's too serious!

However, it suggested using the risk insights to develop design guardrails or “do this, not that” guidelines to inform concept development — and I loved the results!

#aiexperimentlog #1linkedin #highsuccess

https://docs.google.com/document/d/1dS6HRci_fmX5hGC0K3snbkZqAv8cyJjxYcU80Q3S5V4/edit?usp=sharing

🔭 Using HMW questions for concept generation.

I've been skeptical about the utility of ChatGPT for GOOD concept generation, but I was pleasantly surprised in my experiment! These were really well grounded and I think would be a great, low barrier of entry for kick-starting the ol' brain when the creative juices just aren't flowing.

https://docs.google.com/document/d/1v9gniZkSiN-cS1Kiy55lIEeWXkreHboJp-5dbJkHnmc/edit?usp=sharing

#aiexperimentlog #1linkedin #highsuccess

Generating HMW questions and using them for reverse brainstorming. The results were both hilarious and insightful.

I laughed out loud at, "Add a 'feeling lucky' button that fills your Experience section with AI hallucinations."

On the flipside, the "Realistic but Risky" ideas made me think more about long term impact. This could be a lightweight method to "move slow and fix things" in an agile environment.

https://docs.google.com/document/d/1BmZJaO7Lot-T653JvBTWGhb8vnOYNT0JGXgcwUB8X9c/edit?usp=sharing

#aiexperimentlog #1linkedin #highsuccess

Transitioning from DEFINE to IDEATE. In this experiment, I did a retest of Leighann's user story template and worked with the GPT to prioritize them based on product impact.

As a solo designer I might've just picked the one that resonated with me the most. I think this was more objective and made me consider other factors external to user needs/desires.

https://docs.google.com/document/d/1MyHhLw2U0W_Hz7qE2xk0bOLoOnsrLL-Bmwx27Embmvg/edit?usp=sharing

#aiexperimentlog #1linkedin #highsuccess

Retest of my previous experiment, re: combining qualitative and quantitative data.

This time I prompted in stages to better understand what info the LLM needed from me in order to be effective.

I think I was the problem this time! I don't have much experience combining qualitative/quantitative data and I'm realizing if I'd structured my initial research process differently, this could've been more effective.

https://docs.google.com/document/d/19FpfDhUvQnOavKgjeAhRwh3l9CNClG2LrQjHdZuFeZ4/edit?usp=sharing

#aiexperimentlog #1linkedin #medsuccess

Everything, everywhere, all at once! Tried to synthesize insights from quantitative and qualitative data, combine them to extract meaningful insights, and suggest next steps all in one go.

Total fail.

It ended up making little to no connections between the data types and pulled out insights from the interviews that I perceived as less relevant. Overall, the analysis felt very shallow and didn’t reflect the actual data.

#aiexperimentlog #lowsuccess #1linkedin

https://docs.google.com/document/d/1DItrvSt86r435iWlykImf6tOCWdMff_A9Nm9hWH22aA/edit?usp=sharing

Experimenting with survey design to collect quantitative data to inform the Define phase.

I also wanted to test out pushing AI to think a few steps ahead. I tried 2 different prompts and got different results (more general, more specific). I think it's personal preference and/or process-dependent.

I'm enjoying seeing where AI can go, but I also get the icky feeling that it's no longer MY work... am I a designer or a manager??

https://docs.google.com/document/d/1Rr627r3zdn9dZ2GYt-T6pPHsD0rpB2TwDSC33vdA7Dc/edit?usp=sharing

#aiexperimentlog #1linkedin #highsuccess

Using my thread with NN/g's Research Assistant GPT to compare human-derived insights with AI-generated ones.

This gave me the warm and fuzzies — I loved how collaborative it felt! Really struck a balance and I felt like I could trust it more after this process.

https://docs.google.com/document/d/1ljPEBsuXVY4TtV154WOp1n9F99gndRJjIDQpIyxJ7ng/edit?usp=sharing

#highsuccess #1linkedin #aiexperimentlog

Using data gathered by team #1linkedin to compare human-derived insights to AI-generated insights.

My second experiment uses NN/g's Research Assistant GPT. High information fidelity, clear and concise insights, well-documented next steps. It jumped all the way ahead and I trusted the process!

The only limitation is the inability to dynamically link the source data for easy reference.

https://docs.google.com/document/d/1Oa_dQCg-9Bo-1FxeTiVXgmCr4fKWnleyL4L13bOwHF8/edit?usp=sharing

#highsuccess #1linkedin #aiexperimentlog