Chapter 1: Design for behaviour, not screens, lesson 4 of 6

Find what you bring that AI can't

AI can make a persona or a screen in seconds. So what is left for a designer? This lesson gives you the evidence, asks you, and ends with a thought experiment.

By 4 min read, 14 min of exercises

Warm-up

A quick one before we start. Is each one a frustration, a motivation or an aspiration?

  1. I want to be a teacher who knows what each student can do.

    Right, it's aspiration. It's who he wants to be. That's why grading matters to him at all.

    Not quite. It's aspiration. It's who he wants to be. That's why grading matters to him at all.

  2. I need to hand in grades by the end of term.

    Right, it's motivation. It gets him through this term's task. Then it's done.

    Not quite. It's motivation. It gets him through this term's task. Then it's done.

  3. AI wrote my students' code, so I can't tell what they learned.

    Right, it's frustration. It's a problem he has today, and it costs him every time he grades.

    Not quite. It's frustration. It's a problem he has today, and it costs him every time he grades.

AI can now write a persona, a journey map or a set of screens in seconds. So what is left for a designer to do?

I’ll be honest with you. When I was asked this, I said: I struggle with the question of what it is that I bring to the table that AI can’t do. I still don’t have a neat answer. So this lesson won’t hand you one. It gives you the evidence from this chapter, and asks you.

Think first

What does a designer bring that AI can’t?

You

Which of these is closest to your answer?

Shirish

AI gets better at new combinations every month. So be careful with a claim you can’t show.

What would prove that a design was creative, not just new?

Shirish

That’s close to where I landed. But be careful. Much of the evidence in this chapter sat in logs, like the second emails. AI reads logs faster than you do.

Which part of understanding people can’t come from data?

Shirish

I think that’s the strongest answer in this chapter. The teacher admitted something hard, because he trusted me.

But trust takes time. What does a designer have to do, again and again, to earn it?

Shirish

Test that on this chapter’s stories. The teacher told me something he wouldn’t type into a form. He told a person he trusted.

Could a chat window have heard it?

You

What this means

Nobody has a settled answer to this, including me. But you don’t need one to start. You need evidence to reason with, and this chapter has some. Let’s sort it.

What got cheap, and what didn’t

Here are jobs from this chapter. Some, AI does well today. Some, it can’t do on its own.

Practice

Can AI do each job without a person?

  1. A user tells you two things that don't match. Notice it, and ask why.

    Right, it's a job that needs a person. You have to be in the room to notice, and brave enough to ask.

    Not quite. It's a job that needs a person. You have to be in the room to notice, and brave enough to ask.

  2. Count how many emails get sent again with a file a minute later.

    Right, it's a job AI can do. It's a pattern in a log. AI finds these fast.

    Not quite. It's a job AI can do. It's a pattern in a log. AI finds these fast.

  3. Get a user to tell you a hard truth about their work.

    Right, it's a job that needs a person. People share hard truths with someone they trust, not with a form.

    Not quite. It's a job that needs a person. People share hard truths with someone they trust, not with a form.

  4. Turn interview notes into a first draft of a journey map.

    Right, it's a job AI can do. It can draft it quickly. You still check every line against the notes.

    Not quite. It's a job AI can do. It can draft it quickly. You still check every line against the notes.

The pattern: making things got cheap. Knowing what’s true, and earning the trust to hear it, did not.

The teacher who couldn’t grade

Here is the answer to your bet from lesson 3.

Shirishfrom my own product

The teacher asked for help with grading, then said he didn't need it

The same teacher asked me for help with grading. Later he said he was happy grading by hand. The two answers didn’t fit, so I asked him why.

His students had started writing their code with AI. The code worked. But he couldn’t tell what any student actually understood.

So the answer to your bet: knowing what each student understands.

The code got cheap. What got rare is proof that someone understands. That is true for his students, and it’s true for designers too. And notice how I found out. He told me because he trusted me, and because I asked why when his answers didn’t fit.

A thought experiment: the Experience Machine

Thought experiment

The Experience Machine

Robert Nozick, Anarchy, State, and Utopia, 1974

Think first

Imagine a machine that can give you any experience you want. You plug in. For the rest of your life, you feel like you are doing your dream work. You design products people love. Every client is happy. Every launch goes well.

None of it is real. But while you are plugged in, you don’t know that.

Would you plug in? If not, what would be missing?

You

Which of these is closest to your answer?

Shirish

Test that with one detail. In the machine, no client ever disagrees with you. No design ever fails. What would you learn after ten years in there?

Shirish

That is what Nozick thought too. Now go one step further. What does really doing the work give you that the feeling doesn’t? Name one thing.

Shirish

That was my answer too. I like my imperfections, and I want to see myself grow. Notice what this means. Mistakes are not just unpleasant. They show you where you were wrong. And finding that out is how you change.

You

What this means

Nozick used this machine to make a point. We care about more than how life feels. We want to really do things. We want to become a certain kind of person. And we want to be in touch with what is real. A life of perfect feelings in a machine has none of these.

Why does a UX course care? A good feeling and a finished task are two different things. You can design for one and lose the other.

Now in UX

Think back to my food delivery app from lesson 1. In testing, people said they liked how it looked. After launch, very few people used it.

Were the testers in the experience machine? What were they telling us, and what should the test have checked instead?

You

Which of these is closest to your answer?

Shirish

Yes. I like how it looks is a report from inside the machine. So what would you have asked them to do? And what would count as success?

Shirish

That’s possible, and worth checking. But they said they liked it would look the same either way. What proof would tell the two apart?

You

What this means

A compliment in a test, a happy score, eight out of ten: each one tells you how the experience felt. None of them tells you if the person reached their goal. This course keeps asking the other question: what did they do?

And the teacher’s students? AI took away their mistakes. Their mistakes were the proof they were learning. That is the experience machine again, in a classroom.

Words from this lesson

These words come back in the rest of the course.

Artefact
Something UX work produces. AI makes these cheap to make. In this lessonAI can draft a persona or a journey map in seconds.
Judgement
Deciding what is true and what matters. AI can help, but you are responsible for it. In this lessonI noticed the teacher's two answers didn't fit, and asked why.
Trust
What makes people tell you the truth. A form or a chat window rarely earns it. In this lessonThe teacher told me he couldn't grade his class.
The experience machine
A reminder that a good feeling and a finished task are not the same thing. In this lessonTesters said they liked my app, but real people didn't use it.

Back to your first answer

Your first answer

Your answer

Your answer will show here after you write one at the top of this page.

You've read the lesson. Would you change your answer?
Good. Which part of this lesson backs your answer?

Your bet: how do you find out you’re wrong?

Your bet

Chapter 2 starts here. You believe a new feature will help people.

What is the fastest way to find out you’re wrong? Make your bet, and say how sure you are.

Your guess
How sure are you?

You'll find the answer in Lesson 2.1: Find the beliefs behind an idea.

Sources

  • Robert Nozick, Anarchy, State, and Utopia, 1974: the Experience Machine.
  • The teacher story is from my own product.
Portrait of Shirish Shikhrakar

Written by

Shirish Shikhrakar

Shirish is a Google-certified UX designer and UX engineer based in Kathmandu, Nepal. He started as a software developer, moved into product design for Silicon Valley startups, and has taught UX foundations to hundreds of designers since 2019.

AboutLinkedIn