A friend is getting back into UX research after a two-year gap. Recognizing AI is a big deal, she asked where to start learning. Great question! Other people are surely wondering as well, so I’m sharing my answers here.

Caveats:

  • These answers reflect how I learn; YMMV.
  • You’re aiming for employment, so we’ll focus on practical knowledge.
  • I wrote this in early Feb 2026 — it won’t age well.

Finally, you’re behind the curve, so I’ll abandon nuance and be blunt, broad, and brief. Let’s go!

1. Get Hands-on ASAP

Don’t bother with AI-powered products that automate particular tasks. Most are front-ends to specialized prompts, contexts, and data sources. In time, most will be replaced by general-purpose systems.

Learn first principles instead. If nothing else, sign up for a paid Claude or ChatGPT account. Learn to use projects, Custom GPTs, and other such “advanced” features.

But you should quickly go beyond chatbots. This entails coding and (at a minimum) working in your computer’s command-line interface. The CLI is multi-decades old, there are lots of guides online.

Once you’re comfortable with the command line, install Simon Willison’s llm. Play around. Build little automations for everyday tasks. When you’ve “got” llm, install Claude Code. Use it on something larger than a papercut.

Plain text is the lingua franca here; learn to represent data in Markdown and how to structure context. If your computer has enough memory, install a local model. (If it doesn’t, consider getting a more powerful machine.) Explore Hugging Face.

If you get stuck, look in YouTube. Lots of hands-on guides there. And of course, Claude/ChatGPT can be fabulous tutors. Set up a project in either one to support your learning journey.

Self-serving: if you’re attending the IA Conference, register for my AI Hands-on workshop. I occasionally teach an online cohort as well; sign up to my newsletter to be notified.

Bottom line: Start making as quickly as possible. Use AI to solve real problems. Document your experiments to show potential employers. (Here are mine.)

2. Follow the Right People

You need theory too. Alas, the field is changing too fast for books. (With one exception, noted below.) Your best bet is to follow the right people.

“Right” means

  1. they know what they’re talking about,
  2. regularly share useful/insightful stuff, and
  3. aren’t grossly biased for or against the technology.

That sounds stupid, but there’s lots of hucksterism and ideological bloviating around AI.

Here’s who I follow:

  • Simon Willison: the prototypical alpha geek, experimenting hands-on and generously sharing what he learns.

  • Ethan Mollick: a practical academic who deeply understands the tech; his book Co-Intelligence is the best general-purpose intro.

  • Andrej Karpathy: OpenAI co-founder; shares (long!) videos on how the technology works.

  • Gary Marcus: highly critical of the hype; understands the technology’s potential but (rightly) calls out its shortcomings.

I’m skeptical of:

  • Academics in the “soft” sciences
  • Mainstream journalists
  • Anyone else with a megaphone and likely to lose status
  • AI company execs (and others looking to inflate valuations)
  • Doomsday prophets

Yes, there are ethical, environmental, legal, financial, etc. questions. Lots of people have strong opinions on these subjects, but there’s little solid data. You’re looking for practical advice, so I suggest putting aside these concerns for now.

3. Re-think Your Work

As you learn about AI, ruthlessly consider the impact on your work. Look to replace yourself. I assure you, others are. Do it first.

Research is one of the areas of UX that will most be transformed. I guesstimate 80–90% of “traditional” jobs will disappear. New roles will look more like management than traditional IC roles.

Ask yourself:

  • How would I delegate this task to an AI “intern”?
  • What information do they need to do it right?
  • How would I measure results?
  • How would I give them feedback?

Assume the intern has pattern-matching superpowers and knows more about what you’re delegating than you do. What it doesn’t have is humanity and common sense.

That’s where you come in.

Be the Human in the Loop

Research is sense-making: gathering relevant data about a problem space and asking the right questions to generate insights that support good decisions. AI can greatly augment humans in the process. It can’t fully replace them — yet.

But it’s not a question of whether organizations will use AI in UX research. They already are. The question is how they do it. Experienced practitioners will guide them to better insights cheaper and faster, while doing it ethically and in service to human goals. But only those who grok the technology will have a say.

Get cracking.