How to use ChatGPT as a research assistant

 
 

In every design project I’ve undertaken, insights about human needs have formed the foundation for our solutions.

Now, with ChatGPT, we all have access to an intelligent research assistant that can stimulate our thinking and handle much of the legwork. Notably, this research assistant is growing exponentially more intelligent with each release.

Here are four ways I’ve been using ChatGPT (and some other tools) to enhance your research:


1. Grasping the big picture of a theme

When researching an unfamiliar topic, it's helpful first to develop a broad understanding of the subject. ChatGPT excels at this.

Prompt ChatGPT with “Give me an overview of [theme]. Keep things on a high level.” Then follow up with more specific questions. This method is far superior to using Google for learning about a theme. I wrote a more in-depth piece on learning with ChatGPT, which you can find here.

ChatGPT can also recommend literature and other sources like YouTube channels to familiarize yourself with a topic. It can even find the latest scientific findings, but be cautious, as ChatGPT may occasionally provide links to non-existent research findings.

For a more scientific foundation, tools like Consensus can help you understand the scientific consensus on a given topic and connect you to relevant studies more accurately than ChatGPT.

Keep your conversation with ChatGPT in one thread to retain the context as you move on to the following stages.

ChatGPT can help you understand the broad overview of a theme


2. Formulating research questions with ChatGPT

The quality of your research often depends on the thoughtfulness of your questions.

While you should lead this process, tools like ChatGPT can act as valuable assistants.

Start the conversation with a simple prompt like “I’m conducting user research related to heart health. You are an expert on user research. Can you help me figure out what to ask?”

To maximize ChatGPT's usefulness in this context, I recommend:

  • beginning your prompt with a broader context of the theme

  • engaging in an ongoing conversation with ChatGPT, providing feedback, and guiding it in different directions

  • prompting it to be an “expert on user research” for more nuanced answers

  • using ChatGPT to draft both interview and survey questions

  • treating ChatGPT's input as a first draft for inspiration, iterating with your own ideas

  • seeking feedback from ChatGPT on your interview outline to identify potential gaps

ChatGPT can help you write a first draft of interview questions. Don’t take it as a final output, but something to get the gears turning.

3. Conducting a mock interview with ChatGPT

A creative way to use ChatGPT for research is by conducting mock interviews.

Ask ChatGPT to “act as an interview participant. I will interview you for a user research project related to heart health. You are a 60-year old woman from Chicago with a history of heart issues. Answer my questions one at a time. The first question is...”

While this doesn't replace interviewing real people, it can:

  • uncover common themes and challenges based on ChatGPT's learning data

  • provide a trial run to assess the interview flow and identify areas for improvement

Conducting a mock interview with ChatGPT can help you tap into humanity's collective knowledge, but it cant replace the deep nuance and insights from interviewing actual people.

4. Analyzing research data with generative AI

Traditionally, we've relied on our own abilities to synthesize findings from user interviews and surveys.

Now, AI can supplement our pattern-matching capabilities.

Examples of using ChatGPT to analyze research data include:

  • summarizing critical findings from raw notes or voice-to-text transcripts of interviews

  • analyzing open-ended responses from surveys

  • examining messages from public sources like Reddit or Twitter to understand how people discuss a specific theme

Conclusion and the way forward

The more I experiment with ChatGPT, the more I see our "relationship" as a deeply collaborative one throughout knowledge-based tasks.

This holds true for research as well. It's helpful to have a ChatGPT window open to bounce ideas off, analyze data, and discover patterns you might have missed throughout the various stages of research.

It's crucial to consider research participant privacy and confidentiality. Naturally, don't upload any personally identifiable information to ChatGPT or any other tools.

In the coming months and years, the multi-modality of AI tools will continue to advance. This means that within one tool, or a combination of many, we'll be able to analyze and produce text, images, video, and sound. The implications of this progress for user research are undoubtedly transformative.

 
Matias Vaara

I help teams tap into the power of generative AI for design and innovation.

My weekly newsletter, Amplified, shares practical insights on generative AI for design and innovation.

Previous
Previous

How to enhance your communication with generative AI

Next
Next

A strategic framework for generative AI opportunities