We’ve heard the question from many researchers, designers, and product managers: Where is a good place to strategically use AI in their work? And what are the pitfalls they should be aware of?

In “The New UX Tech Stack: How to Strategically Use AI”, Laurel Brown (Senior UX Researcher at Dscout) and Nikki Anderson (Founder of User Research Strategist) dove into the details. You can watch the webinar here. 

Below, we’re highlighting the suggestions they provided, with examples you can start using right away. 

Using AI as a thought partner

Instead of automatically integrating AI into important workflows, start with low stakes experiments to get a feel for how the tools work. 

For example, test it on personal tasks like asking AI to draft recipes with whatever is in your fridge or help explain how to play a new sport. Even if the answers aren’t perfect, you’ll learn how the tool makes mistakes—and how to spot them. This is also a great opportunity to see how AI structures options and ask clarifying questions.

The advantage of this approach is reframing AI as a process of discovery instead of an overwhelming challenge.

Once you’re comfortable with how AI behaves in low-stakes contexts, it’s easier to transition into professional ones. The same principles (test, explore, and learn where it breaks) apply to research. 

By treating AI like a curious but inexperienced partner, you’ll gain confidence in directing it, setting boundaries, and eventually using it for different research tasks.

Prompts you can use today

Speak stakeholders’ language

AI can help you cultivate stakeholder empathy and act as a research partner. One approach? Drop in your research plan or report and ask AI to critique it from the perspective of a skeptical stakeholder.

Use prompts like:

• “Critique this research plan as a skeptical VP of Product.”
• “What’s irrelevant, unclear, or missing in this report?”
• “What would make this seem lacking or out of touch?”

Vetting these questions reduces your chance of being caught off guard in a big presentation, for example. By surfacing weaknesses early on, you’ll also make insights more relevant and accessible to the people who rely on them.

Prove business value 

Research and design teams often need to frame findings and outcomes in terms of business value. This ensures that your research will be valued down the road, and that you’ll get continued resources and recognition. 

To orient around business value, ask AI to reflect on the potential perspectives of leaders:

• “For the head of product in this situation, what business metrics would matter?”
• “What company/team goals might be relevant to this study?”
• “How could this study help the team move forward with their decision?”

This helps create study plans or proposals that clearly link insights to organizational outcomes.

“My proposal has to prove that there is some sort of value that they’re getting from bringing me in. That is really important for me.”

Nikki Anderson-Stanier
Founder, User Research Strategist

Test your plan and assess risk

Context makes all the difference. Provide AI with details about your stakeholder or audience, then provide specific prompts. 

For example, feed in:

• The stakeholder’s past feedback style
• Their role and responsibilities
• Examples of what they’ve dismissed or challenged before

Then, you can ask prompts like:

• “What feels vague, low impact, or like a poor use of resources?”
• “Help me spot weak points in this study plan now: where might I get stakeholder pushback, or where is the framing off?”

Frameworks

The clarity chain

When stakeholder requests are vague, consider using AI to create more clarity. You can do this by iterating through stages:

1. Drop the ask: Paste in the vague request.
2. Follow up: Ask AI to reframe it into at least one clear research question, plus suggested follow-up questions.
3. Expand options: Request multiple scoping approaches (e.g., “Give me three ways I could scope this.”).
4. Final pass: Generate a one-paragraph framing suitable for a kickoff document.
   

Don’t automatically trust what AI tells you. Always add guardrails about where it should pull its information from—especially when it comes to cited sources. Follow up by asking if it is hallucinating or making something up, and cross check any references or citations it provides.  

Research pre-mortem

Before you go ahead with running a study, consider using AI to simulate how that study could fail.

Prompts to try:

• “What’s the worst part of this study design?”
• “Where might this plan fail?”
• “What makes this feel like a check-the-box exercise?”
• “What’s one missing question that would shift the value of this research?”

This process reduces risk, and helps you keep an eye out for potential pitfalls ahead. 

Participant mirror

Bias can creep in through recruitment or study design. AI can help you surface what you may be missing.

Prompts to try:

• “Who am I not hearing from who might challenge assumptions?”
• “What kind of participants am I unintentionally filtering out?”
• “What’s missing from the way these questions are framed?”

Remember that AI is not an expert, and sometimes it replicates pre-existing biases from other sources. Try out using its responses as brainstorming material, but not as directives.

AI in your workflow

Planning and recruiting

AI can be a great thought partner and tool to spark ideas during early phases. As always, human discernment is key here. Don’t immediately take everything AI says at face value. 

Use AI to:

• Sharpen focus
• Draft questions 
• Audit for bias and clarity
  

Be discerning about:

• Strategic decisions
• Contextual criteria
• Generic solutions 

Building and fielding

AI can help with simplifying your task flows, but also be cautious about using it to source your field questions or do research for you. Always cross-verify any research AI provides to you, and be cognizant about what kind of data and information is confidential or sensitive to share. 

Use AI to:

• Simplify existing task flows
• Clarify instructions
  

Be discerning about:

• Data collection 
• Crafting test flows
• Nuances participants may encounter that AI cannot predict

Analyzing and reporting

AI can be incredibly helpful in the initial analysis phase, but use caution when it comes to synthesis and determining the research findings. Feeding confidential or proprietary information into AI may also come with legal risks, so always check with your legal team what information is allowed to be shared on that front. 

Use AI for:

• Initial analysis 
• Refining insights
• Tailoring output
  

Be discerning about:

• Strategic decisions
• Complex synthesis
• Nuanced findings
  

“Always check with your legal team before you do anything. It’s so important. You can’t just throw stuff into AI.”

Nikki Anderson-Stanier
Founder, User Research Strategist

Wrapping it up

When using AI in your research process, it’s important to…

• Use your critical thinking skills
• Discern which prompts and tasks are actually fitting for AI
• Follow up on AI’s potential to “hallucinate” or lie
• Cross-check answers or research for verification 

For the time being, AI is typically not best suited to drawing up research plans from scratch or providing nuanced analysis. However, it can be a great partner to…

• Brainstorm ideas
• Find blind spots in your proposals or presentations
• Gut check for biases
• Tie your work to stakeholders’ needs and business outcomes 

As your use of AI grows and evolves, you’ll also get a better sense of when it’s best to apply it. At the end of the day, keeping humans front and center is at the heart of any quality research work. 

You may also like…

• Roundup: Your Multi-Use Starter Pack for Applying GenAI to UX
• How to Create Meaningful Personalization in Your AI Products
• AI Heralds a Seismic Shift in Design. Here’s How to Evolve