“I estimate that I saved 20-30% time”

Hi Johan Salo! You’re a Service Design and CX consultant at Ambition in Malmö. You’ve been using machine learning and AI in your work since 2019. Most of us didn’t discover AI as a tool for research and design until just recently. What were you doing back then, and how has your work changed over the years?

– Absolutely, I started out by exploring GPT-2 and how it could be used in research. Back then the technology wasn’t quite there, but you could certainly get a sense of where we were going. It was obvious that this technology would affect not just research, but all kinds of work.

– My role has changed quite a bit over the last couple of years. Today I use different AI tools to analyze large amounts of data and find correlations that may be difficult to identify through more analogue approaches. I see myself as a curator of different tools and methods, and use AI where necessary in various tasks.

Interesting. So can you give me specific examples of projects where generative AI has been significant?

– Of course. A while back, I was conducting a large survey study to evaluate and improve a customer loyalty program. I prompted GPT-4 to give me a draft based on the goals of the program. I then edited and adjusted the draft, and iterated it a few times with GPT-4. I then ran a trial study, evaluated the answers by myself and with GPT-4, and made a new round of adjustments before running the full-scale study. Compared to similar work I’ve done before I estimate that I saved 20-30% time, and I’m also sure that the survey itself was qualitatively better than my previous work.

– In these cases, I want to stress the importance of using solutions like OpenAI ChatGPT Enterprise with high security and privacy, and where the data you use as input is not used to train the model itself.

What advice would you give to service designers and UX researchers who are interested in exploring AI tools as partners to speed up and improve their work?

– Experiment. ChatGPT or Gemini are excellent starting points. Prompt them with what you’re trying to accomplish and ask them to provide you with e.g. a research plan, or a survey. Ensure that the context and content you provide are thoroughly anonymized unless you’re using secure services (i.e. not the open versions of ChatGPT and Gemini).

– Also, I’d be happy to connect here on LinkedIn. Just let me know if there’s anything I can help out with.

Thank you, Johan. Both for this chat and for the tips I’ve gotten over the years. I’m always grateful for the tidbits you send my way.

Sign up for our newsletter

We help designers and innovative organizations to grow – both in capability and confidence. Sounds interesting? Sign up for our newsletter, and let’s keep in touch.

Discover more from Ambition

Subscribe now to keep reading and get access to the full archive.

Continue reading