Ever since OpenAI burst onto the scene at the end of 2022 and made a little known chatbot a household name, discussions around AI have been heated to say the least. From last year’s writers’ strikes to the renewed uproar over the announcement of Sora, creatives have consistently been some of AI’s most vocal naysayers. 

Some may say with very good reason.

‍

‍

After all, many have struggled for years to carve a tenable living within highly competitive industries; the notion of being replaced now by a robot is unconscionable. And that’s before you even delve into the very murky field of AI copyright. There’s huge debate over the ownership of the imagery that AI has been trained on, with opponents arguing that it doesn’t create anything of its own, rather patches together that which has already been created by uncompensated human hands.

Yet there’s one key element that people have been overlooking in all of these debates. Something that AI cannot recreate: taste, aesthetic sensibility, creative discernment.

As AI-generated imagery makes it possible for anyone to create even the most surreal scenes, like a gorilla on the moon, a question arises: does this make one an artist? The answer lies not just in the creation of bizarre or striking visuals, but in the ability to convey deeper meaning about human experience or artistry itself. Without this underlying substance, such images may not truly contribute to the world of art.

In other words, just because you can generate an image of a gorilla on the moon it doesn’t mean that you should. Art is about more than merely conjuring realistic depictions, it has to move people, it has to have something to say. The proliferation of AI tools has democratized technical skills, and that makes good taste the distinguishing factor in art.

But then, what is taste? Is it just personal opinion? Or is it something external, something measurable? 

In a fascinating essay, Paul Graham attempts to chart the rules of good taste from mathematics to architecture and film. He proves that there are universal laws that make it possible to discern whether something has merit, whether liking it is in good taste.

Metrics were measured across multiple industries to chart patterns across what people generally agree to be ‘good’. And he found the following tenets to be true:

1. Good design is simple
2. Good design is timeless
3. Good design solves the right problem
4. Good design is suggestive
5. Good design is often slightly funny
6. Good design is hard
7. Good design looks easy
8. Good design uses symmetry
9. Good design resembles nature
10. Good design is redesign
11. Good design can copy
12. Good design is often strange
13. Good design happens in chunks
14. Good design is often daring.

‍

People often say that good taste is subjective because it’s easier to say that. It’s the inoffensive thing to say. 

But then why do we have critics? Why bother having any discourse around art or culture at all? What Graham proves is that there are standards. There are objective metrics which it is possible to use to assess a piece of work or compare it to other pieces of work.

It will be more important than ever to use these metrics in the age of generative AI. If we say everything is art and give anything a free pass, we’ll be inundated with ‘art’ to the point of meaninglessness. It’s precisely because the barriers to entry have been lowered that we must uphold our creators to more stringent critical standards.

Graham begins his piece by sharing an anecdote. His friend, a teacher at MIT, used to assess graduate student applications. "A lot of them seem smart," he said. "What I can't tell is whether they have any kind of taste." You see, it’s not enough to possess the technical smarts if you have no idea how to put them to use in a tasteful way. And that’s precisely what people are missing with AI. 

It was never just about technical aptitude. Anyone can learn to wield a paintbrush and even become a competent painter, but not everyone can become Picasso just because they have access to an easel. It is the human behind the tools who imbues the output with significance. In the right hands, generative AI could be a tool like any other, one that allows artists to express their point of view and share their unique take on the world. In the wrong hands, you get dogs on skateboards and gorillas on the moon. 

‍

‍

Looking at the work of the all time greats in their field is a way to understand good taste. Attempting to examine something with a critical eye and understand why other people feel it’s objectively ‘good’ is also an important part of developing taste.

Once you’ve absorbed enough perspectives and facets of culture, it’s then time to turn inward. Begin to interrogate yourself. How does this piece of art make me feel? What is this novel trying to say about the world? What was the artist, designer or mathematician trying to achieve when they created this, and how successful were they in doing it?

These questions will become more important than ever as AI continues to evolve. The real challenge for artists will not be using these tools, but becoming clearer on defining their own taste, the very thing that drives their practice. 

‍

‍

Personal taste tells the story of who we are. In choosing what books we like or songs we listen to, we’re essentially saying - this is who I am. This is what speaks to my soul. Good or bad, our taste reveals something about us. That awful film you love tells the story of that time you had chickenpox, and that strange sandwich combination might conjure up memories of your grandmother. These little quirks in our taste, our guilty pleasures, are endearing and are what makes us uniquely human. Generative AI, without any of those experiences, cannot replicate that.