Utilizing ChatGPT For Research Synthesis

Leveraging AI for research synthesis

I’ve always enjoyed planning and facilitating user research as a designer. The past few companies I’ve worked for didn’t have dedicated user researchers so it was up to the design team to get those learnings. A part of the research process I didn’t enjoy as much was synthesizing the research to connect the research calls and summarize the main learnings and figure out which areas to further investigate. Thanks to ChatGPT this process has been a lot smoother.

My Current Process

At my current company we use Zoom for remote research calls that get uploaded to Gong. From there I take the transcript and upload a portion of it a time. The reason for this is because ChatGPT has a character limit when uploading text and I actually like to not rely completely on ChatGPT. I find it helpful to also attempt to synthesize the main learnings while using ChatGPT to help me.

Note: The GPT model I am currently using is GPT-4

I ask ChatGPT to summarize the transcript with the following prompt.

“Can you please summarize the main learnings from the following research transcript?” [Insert research transcript here]

Note: Are any of you saying “please” when asking ChatGPT, it’s kind of funny but I want to be kind when asking ChatGPT lol.

From there I review the summaries to make sure it makes sense and I copy and paste the summaries over to a consolidated Google doc with all my research learnings.

Company Shareouts

I then pick 2 to 3 main learnings from the research call and then share them in my company Slack channel. I only include the 2 to 3 main learnings in the Slack message and link to the Google doc and research recording if anyone wants to learn more.

Diving Deeper

The impressive thing about ChatGPT is that it also helps to use it as a way to get help with synthesizing because if questions come up or if there are topics you want to learn more about you ask ChatGPT to learn more about those topics. By learning about topics that come up during the research calls you end up learning more in my opinion compared to just summarizing the call the old fashioned way.

Prompt Engineering

At times it also helps to get the counterview point regarding something that was mentioned during the research call. You can ask ChatGPT to provide the counterview point and have a conversation with it to help you think things through more holistically which is something I recommend trying. If there is something that resonates with you but you want to have a conversation to learn about the counterview points then try ask ChatGPT something like “For the following responses can you please provide the counterviewpoints in the writing style of “insert a famous author or thinker”. I’ve found the responses to be interesting so far.

Next Steps

It’s early but so far utilizing ChatGPT for research synthesis has helped me save a lot of time and be more productive. I’ll share more learnings as time goes on. If any of you have any questions reach out to me on Twitter.

Thanks for reading.