You still need the grunt work

I was recently invited to The Ken’s podcast to discuss an interesting question: if AI takes over the grunt work through which early career designers traditionally learned, how do they build judgement?

I spent quite a bit of time thinking about this while preparing for the conversation, so I thought I’d share some of those notes here.

My short answer is: you still need the grunt work. But the grunt work is changing.

You can’t jump to judgement

For the longest time, the learning loop in design, or really any craft, was fairly straightforward.

If you were an early career designer or engineer pre-2022, you mostly did the work nobody senior wanted to do. You made the fourth variation. You resized the banner. You took the first pass at something.

Someone more experienced looked at it, told you what was wrong, and you went back and tried again. Do that a few hundred times and, somewhere along the way, you start developing judgement. AI messes with this loop because it is getting really good at the first part.

It can generate the variations, resize things, write copy, explore directions and increasingly build the interface itself. So can you just skip all of that and jump straight to the judgement layer? I’m not so sure.

The analogy I keep coming back to is cooking. You can eat at great restaurants, read recipes and have lots of opinions about food. But if you don’t actually cook, those opinions only take you so far.

You build taste by cooking repeatedly. You serve people. You notice what they like. You screw things up. You figure out why. Then you cook again.

Design works the same way. AI can compress the journey quite dramatically. I don’t think it can eliminate the need for experience.

Build things that have consequences

This is also why I’m not a huge fan of hypothetical portfolio projects. There has probably never been a better time for an early career designer to just make something real.

Build yourself a meal planner. Make a tiny music app. Help an early stage startup. Work with a non-profit. Fix something that annoys you. Make something for your parents.

The idea really doesn’t matter that much.

What matters is going through the whole loop:

1. Find a real problem.
2. Make something.
3. Put it in front of people.
4. See what happens.
5. Get feedback.
6. Improve it.

And ideally, find ways to increase the stakes.

Maybe somebody is paying you for it. Maybe people use it every day. Maybe a business actually depends on it working. Because that changes how you make decisions. Judgement gets built when decisions have consequences.

Without that, it’s quite easy for “taste” to just become a collection of opinions about what looks good.

Move closer to decisions

The other thing I’d do is deliberately move closer to decision making. If execution is getting cheaper, being great at execution alone is probably not where you want to stop.

Go talk to users. Understand why the thing needs to exist in the first place. Learn how the business works. Get comfortable defending a point of view with judgement as well as data.

Try to help decide what should get built, not just what it should look like.

AI can give you twenty options in a few seconds. But somebody still has to decide which option makes sense, for whom, and why.

And somebody has to be accountable when that decision turns out to be wrong. This is something I look for when hiring too.

Of course I care about the quality of the work. But I’m also looking at what someone chose to show me, how they talk about their decisions, whether they initiate things themselves, and whether they’ve been exposed to enough real-world situations to have developed some instincts.

I care far less about which college they went to or whether they have a Masters. If you’ve actually shipped something, put it in front of users, watched what happened and iterated on it, you already stand out. It tells me you probably won’t sit around waiting for someone to hand you the perfect brief.

DIY your apprenticeship

This is where things get harder. The system for learning this way is, frankly, a bit broken.

If you’re 22, where exactly do you find the senior designer who is going to patiently look at your work every week and help you get better?

For now, I think you have to DIY some of it.

Work on projects. Put your work out there. Find people whose judgement you respect and ask them specific questions. Write about what you’re learning. Find informal circles where people are willing to critique your work.

And keep doing it.

My own journey was pretty much this. I started ownpath because I wanted to create a learning space where I could learn about things I was interested in, especially design and philosophy.

There wasn’t some grand plan behind it. I organised a few small workshops. Then I ran a mentorship program for literally one person. We eventually helped her find a freelance project that paid her about 3X what she had spent on the program.

I also kept asking people for advice. Sidu Ponnappa, for instance, was kind enough to take me under his wing. He eventually became an advisor and investor, but at the time that relationship led to an opportunity to work with Gojek on an upskilling program for their design team.

That led to a product design program. Then Collective. And eventually ownpath Studios. I spent about eight years before that designing and building websites and web apps too.

Looking back, a lot of what I now call judgement around design, technology, hiring or delivering projects came from that combination of doing fairly unglamorous work, asking people for help, getting feedback, and occasionally getting lucky with the people I met.

So I’m a big believer in finding mentors. This doesn’t necessarily mean finding someone who formally agrees to “mentor” you either. A handful of conversations with someone good, around actual work you’re doing, can go a long way. I’ve written separately about how to find a mentor.

I’d also be quite deliberate about where you work early in your career. I actually think design studios are one of the best places for associate designers to start. The incentives are unusually well aligned. A studio can hire someone promising, help them get better, expose them to different industries and problems, and eventually bill their time at a higher rate.

Early stage startups are another good option. They often can’t afford a bunch of specialists, so you end up closer to the founders, users, product decisions and the business itself.

Basically, don’t wait for somebody to hire you before you start doing the job. Start behaving like you already have one.

Design education needs better simulators

DIY apprenticeship is a reasonable short-term answer.

I don’t think it’s a particularly good long-term answer.

We need to get much better at designing environments where people can build judgement before they are expected to exercise it professionally.

I find the Indian Army quite interesting in this context.

You have people in their early twenties leading contingents, running operations and making decisions with very real consequences. Nobody assumes a 22-year-old magically has the judgement to do that. They are put through a system designed to build it.

Aviation has flight simulators. Medicine has surgical training and residency. Germany has a different version of the same idea. Around 1.3 million people go through the Ausbildung system, where they typically spend three years learning a trade through a combination of instruction and actual work. In a sense, they have formalised the grunt work to judgement loop.

Design education in India has very few equivalents. I think that’s the gap.

A good design program should feel much more like a simulator for professional practice. You should have to work with real users, deal with stakeholders, make trade-offs, defend decisions and see what happens when those decisions meet reality.

We built a lot of this into the Product Design Fellowship at ownpath. We had practitioners as mentors and put a lot of emphasis on learning by doing. A big part of the program was industry exposure, either through industry projects or actual freelance work.

At one point, we even got a project from MIT CSAIL and had one of our Fellows work on it.

ownpath Studios itself was born in a similar way. We got an opportunity to work with Philips, and rather than separating “learning” from “work”, we put together a team involving a mentor and Fellows from the Academy to execute the project.

That, to me, is much closer to what design education needs.

Not another perfectly documented hypothetical case study. Not just producing more screens. Real situations. Real constraints. Experienced people around you. Feedback. Consequences. Because pure execution is getting automated by the day.

The goal has to be building judgement through consequences.

Until our institutions catch up, early career designers are going to have to create some of these loops for themselves. The bar has gone up. But I think the tools have also made it possible to learn much faster than before.

So do the work. Ship it. Find people who’ll tell you where you’re wrong. Then do it again. Study the greats. The failures. That is still how judgement gets built.