What Can't AI Do in Design in 2026


What Can't AI Do in Design in 2026

By Hannah Baker


Dear Reader,

If you work in design, your feeds are probably saying the same two things on repeat:

  • Here’s everything AI can do for you, and
  • Here’s why you should be terrified.

Most of that conversation focuses on tools and job titles: “Will designers be replaced?” “Which roles are safe?” It makes for good headlines, but it’s not how the work actually changes in real life.

A few months ago, walking to my studio listening to a Planet Money podcast episode, I realized we’re asking the wrong question.

The episode interviewed two researchers who both focused on tasks rather than jobs, each from a different angle:

  • An exposure rubric: how reachable a task is for AI
  • The EPOCH framework: how much human depth a task requires

I made a modern task list for product/UX design, ran those tasks through both lenses, and then plotted them. I expected a neat line from “very human” to “very automatable.”

That is not what showed up.

Instead, three clusters emerged:

  • Human-core work that stays stubbornly resistant to automation
  • LLM-ready work that’s genuinely accelerable
  • Tool-entangled work where LLM needs the right tools to work

Underneath those clusters lay a quieter pattern: the moments where we stop asking “Can AI do this?” and start asking “What should we do here?”

Not execution. Not output.

That in-between space where context, ethics, and consequences live.

In other words: judgment.

This new report, The Judgment Gap: Design Skills You'll Need in 2026, follows that thread: from a commute to task lists and scatterplots, to why framing, interpretation, facilitation, and ethical awareness are becoming the real craft of design.

If you’ve been half-excited, half-exhausted by the “AI + design” conversation, this is my attempt to offer a different lens.

Facilitating Workshops Cohort 9, is officially happening.

The next cohort dates are March 25-April 29, 2026, and enrollment opens Monday, December 15.

If you’ve been waiting for the next round (or thinking about leveling up your workshop skills for 2026), now’s the time. A few people have already grabbed pre-sale spots this week, so I expect seats to move once it opens publicly.

You can join the waitlist here to get the link the moment doors open.
→ Add me to the Waitlist


FREE EVENT: Upstream of the Machine
Learn how to own the future in this 45-minute masterclass
Sep. 16, 2026 @ 19:00 CEST
Register for Event


COURSE: Defining UX Strategy
Learn to design a winning strategy that aligns design with business.
Next cohort: Oct. 5-Nov. 16, 2026
Buy a Seat ...enrollment is finally open!


Until next week!

Hannah Baker
Facilitator & Co-Founder
The Fountain Institute

The Fountain Institute

The Fountain Institute is an independent online school that teaches advanced UX & product skills.

Read more from The Fountain Institute

Written by a human, not by AI Dear Reader, Last week, some of you were gracious enough to give me some feedback on the newsletter. (You can still do it at the bottom). Someone left a review that said, "AI Slop" without any other context. And it got me thinking... You might be assuming that this newsletter is written by AI. Today, I'll let you in on a little secret: I still write by hand. And I plan to keep writing this newsletter by hand. I'm not gonna lie, I tried to get AI to write the...

Figma AI Skill Library, Save Tokens with Caveman, and the Million Dollar Toilet Dear Reader, A lot happened these last 2 weeks while you were relaxing on the beach. Let's get you caught up! The new Figma AI skills library Figma launched an AI Skills Library. Discover these skills (guided prompts you can use in agents) to unlock new ways of working from research to handoff. Check it out here to see what the community is making with Figma AI. Get your AI to speak like a caveman Caveman.so is...

IBM Deep Blue vs. the chessmaster

The Centaur Strategy Dear Reader, Garry Kasparov's hands begin to sweat. Garry is the world's greatest chess player, but he has run out of moves. His opponent sits there, quietly waiting for Garry to hand him the victory. Garry realizes it's over...not only is he going to lose this game, chess has just changed forever. That's because Garry's opponent was an AI. Garry struggling, from Brittanica On May 11, 1997, the world's leading chess player got beat by IBM's Deep Blue chess program, and it...