#54. Taste and judgement in a fast design process

On Monday, I was listening to an interview with Brian Chesky, co-founder of Airbnb. Among many other interesting things, he said his company now ships features 80% faster than before, because of AI.

That made me stop and think: at this rate, it almost doesn't matter if a feature works. Shipping is the whole game now, and the feature that doesn't perform just gets removed. What about taste and judgement in the process of creating and choosing?

We've been talking a lot, lately, about what AI can and can't take from us. Two words keep coming up: taste and judgement. And I admit I have used them a lot. But are they still a thing in the design process?

A moment of the interview by Silicon Valley Girl (aka Marina Mogilko). Brian Chesky illustrating the many features released by AirBnb lately.

Taste isn't something you're born with

Philosophers have long treated taste as a skill, not a gift. David Hume, in his 1757 essay Of the Standard of Taste, argued that a true judge of taste has a delicate sensibility, sharpened by practice, refined by comparison, and cleared of prejudice. Kant, in the Critique of Judgment, added that a judgement of taste is personal, yet claims to speak for everyone.

I agree with both definitions, because I see it in my daily life.

Everybody compliments me on my “natural" taste: my eye for colour, for putting an outfit together. The same goes for my studio, which feels like a continuation of my own colourful world.
 But I wasn't born with it.
I have been exposed to it since I was a child. My grandfather was a painter, my grandmother a seamstress. The house where my grandparents lived and I grew up was like something out of the colour combination book of Wanda Sanzo.

At 14, I didn't have taste. In my twenties, I knew what I wanted but didn't have the money to express it, so I trained my eyes on the fashion windows of Milan. Taste was still part of who I was becoming, since childhood, but it was built slowly, through continuous exposure and experience. 
Taste changes with context and it needs time to grow in you.

And time is exactly what AI, and the way we build now, doesn't give us.

Sanzo Wada books. These books are based on his original 6-volume work from the 1930s.

Taste, flattened

AI today tends to produce tasteful design based on the average. Anything too different, too quirky, gets discarded fast. That's what taste starts to look like under algorithms. It gets worse, because these systems are trained on our taste but reduce our ability to explore, since the apps themselves are built to engage us and sell more.

Which is why we need to train our taste even harder now, seeking out comparison and direct experience.

I recently came across Fugglers, the deliberately ugly monster toys. I didn't know the brand until a friend showed me a presentation she was working on for it. My first reaction was: oh, that's ugly. Then I realised they're made to be ugly, on purpose. And that stopped me. Maybe not everything needs to be polished and sweet. It's a small counter-example to the idea that taste only moves toward "perfect". It is proof that taste can still push in new directions, against the grain of what an engagement algorithm would ever produce on its own.

That's what taste really is: intention. Someone chose that ugliness, and the choosing is the point.

Fuggler. The tagline of this brand is “Adopt at your own risk”.

Judgement is the other word

Judgement is the ability to make an ethical, moral decision about which direction to pursue. It matters because someone with judgement has a better sense of what consequences an idea might bring.
I have already explored the idea of consequences and machines in my post dedicated to Stanislav Petrov.

Aristotle had a word for this. He called it phronesis, or practical wisdom: the ability to deliberate well about what is good to do in a particular situation, which he held can only grow through experience.

But when we ship features this fast, the friction of ideating and creating goes away, and judgement risks going with it.

This is why, in my upcoming book, I keep insisting on running ideas and products through three realities: people, technology and context, and through time (past, present and future).

Asking how an idea affects people, how the technology behaves, and what context it lands in, and how all three change over time, is slow work. It is the very work that speed skips.

Judgement is also contextual: it may not matter much in something like an Airbnb feature that suggests and classifies the best places and experiences in a given town. But it matters enormously when AI is used in health or in law. The consequences there are bigger than a bad listing.

AI-powered trip planning. AirBnb will soon bring AI into every step of a trip.

Final thoughts

AI is removing the friction we need to develop and refine both taste and judgement.

That friction, the hard and uncomfortable feeling of making something the slow way, was never just an obstacle. It was time: time to ask whether this was good and fit the audience, and whether it opened a new interpretation of good, or of beauty. It was also time to ask whether our effort was well spent, and what it would really cost, not only in tokens burned, but in physical and mental effort, and in consequences.

Time and friction are what AI is removing, and they are also what trains our taste and sharpens our judgement.

Keep an eye on both. Don't cut corners by skipping the friction.

A feature that fails can be removed overnight. A way of thinking, once lost, is much harder to bring back.