Every few weeks the same question does another lap around the internet: is AI going to kill art?
Musicians are asking it. Designers are asking it. The Guardian ran a full essay on it. Your group chat has probably debated it twice this month.
So let me give you the short answer up front. No. AI is not going to kill art.
But something else is happening while everyone argues about the wrong question. AI is quietly splitting every creative field into two groups. People who can only generate. And people who can draw, design and see. One of those groups is about to be worth nothing. The other is about to become the most expensive talent on the market.
Which group you land in is a choice you are making right now, today, with how you spend your practice hours. Let me show you why.
The panic is real, and it's not stupid
In 2024 the Guardian published a piece by musician and writer Rudi Zygadlo titled "Should artists be terrified of AI replacing them?" It opens with him typing his own creative dread into an image generator and getting back something that looks like a holiday ad. Then he says the line that every working creative has felt in their gut: "AI has killed my bucket list."
Why write the album, the screenplay, the novel, when a model can spit out a passable version before your coffee cools? Why spend ten years learning to paint when a prompt takes ten seconds?
If you've felt that vertigo, you're not weak and you're not a luddite. You're paying attention. Every previous tool, from the fountain pen to the word processor, was something an artist held. As Zygadlo points out, AI is different because it threatens to be both the tool and the artist.
So he did something smart. He took the question to the artists who live deepest inside the machine.
The verdict from the people who built art with AI first
Holly Herndon and Mat Dryhurst have been working with AI, and teaching people about it, for more than a decade. They trained an AI "baby" called Spawn to sing on Herndon's 2019 album Proto. If anyone had a motive to announce the death of the human artist, it's the two people who got there early.
Zygadlo asked them point blank: is AI going to kill the human artist? Their answer started with one word. No. Then they gave the reason, in the most clarifying sentence of the whole debate:
"A media file that sounds like a choir is not a choir. Culture will persist and evolve in unexpected ways."
The file is not the thing, and the output is not the art. Their proof is something you've already paid money for: DJing. On a technical level, DJing is very easy to automate. Software can beat-match for you. And yet people pack clubs and drop hundreds on festival tickets to watch a person do it. Nobody was ever buying the audio. They were buying the human in the room and that human's taste.
Writer Luke Fewtrell landed in the same place in his Medium essay "Will AI Kill Art", and he compressed it into seven words: "Art only dies when the artist dies." Humans have been making art since we were grinding pigment onto cave walls, and every technology panic since has followed the same script. The camera was supposed to kill painting. Painting is still here. Synthesisers were supposed to kill musicians. The Buggles sang "Video Killed the Radio Star" and radio is still on air. Art survives every time because people need to make it, whether or not anyone pays. As Fewtrell puts it, the feeling of creating something born from your own mind is unlike anything else.
So the comfortable take is true. Art will survive. But I need you to hear the uncomfortable part, because "art will survive" is not the same as "your career will survive."
AI doesn't kill art. It kills the average.
Generative models are trained on a huge pile of what already exists, and they produce a statistical blend of it. That makes them machines for manufacturing the middle. The output is polished and instantly familiar. It is also interchangeable.
Now apply first-grade economics. When anyone can generate a decent illustration in ten seconds for next to nothing, what is a decent illustration worth? About what it costs to make. The floor of the creative market, the "pretty good and fast" tier, is being priced toward zero as you read this. I wrote about what that sameness does to brands in my breakdown of Meta's AI product photos, and the internet already has a name for the flood: AI slop. People scroll past it on instinct.
So AI is coming for average art. It's coming for the person whose entire value was "I can produce something decent-looking on demand." That skill is now a utility, like tap water.
Which leaves one thing to work out: what the market pays for when decent costs nothing.
Why the pencil just became a power tool
It pays for the things a prompt cannot produce. And every one of them comes from old-fashioned, unglamorous, hand-earned craft.
Drawing teaches you to see. When you learn to draw, the drawing is almost a byproduct. The real product is your eye. You learn why a composition feels balanced, why light behaves the way it does, why a face reads as trustworthy or unsettling. That trained eye separates a person who generates 200 mediocre images from a person who can look at 200 images, know instantly which one works, and fix it. AI raises the value of taste while gutting the value of execution. Taste comes from reps. There is no shortcut, and now the shortcut everyone else is taking is the reason your reps are worth more.
Skills you outsource are skills you lose. Fewtrell makes a point in his essay that most people skim past: over-reliance on AI erodes the thinking it replaces. Creative muscles are muscles. If the machine does every rep for you, don't be shocked when you can't lift anything on your own. The designers who never learn foundations aren't becoming "AI-augmented creatives." They're becoming people who press a button and hope, with no ability to judge or repair what comes out.
Proof of human is becoming a luxury good. In the US, vinyl outsold CDs 46.8 million units to 29.5 million in 2025. Live Nation grew revenue 9% to US$25.2 billion in 2025, with a record 159 million people at its shows, in the age of infinite streaming. The pattern is old: the moment something becomes infinitely reproducible, the scarce version becomes a status symbol. I'd put a hand-drawn identity and a real illustration style in that scarce tier of visual work now. In the Guardian piece, artist Rachel Maclean is already there. While everyone else waits for the models to stop generating eight-fingered hands, she's deliberately preserving the broken models because the errors are the interesting part. That's an artist using AI, not an artist replaced by it. She has the craft to know which mistakes are beautiful.
The window is open right now
Herndon and Dryhurst offer one warning in that Guardian piece, and you should take it seriously. As the tools flatten technical skill, they predict the creative industry drifts toward a popularity contest, where "the most attractive kid in class" will have the tools to choose to be the most popular metal artist or crime thriller author. When everyone has the same generator, attention flows to whoever already has an audience.
I can see only two defences against that world. Be famous already. Or make work that is visibly, provably yours. Most of us don't get option one. Option two is learnable, and my bet is that too few people are learning it, which is why it should pay so well.
Think about the supply curve for a second. Every junior who skips fundamentals and goes straight to prompting is one fewer person who can draw, paint and design from first principles. If demand for distinctly human work climbs while that pipeline shrinks, you don't need an economics degree to see what happens to the price of the skill.
So here's the plan
If you're a creative, or you're raising one, or you're building a brand that needs to stand out in a feed full of synthetic sameness, do this:
- Learn the fundamentals with your hands. Drawing, typography, colour, composition. Analog first. The struggle is the skill, so don't engineer it away.
- Use AI like an intern, never like an artist. Let it explore variations, mock up throwaways, handle grunt work. Never let it make the decisions that define how your work looks and feels.
- Show your process in public. Sketches, drafts, worksheets, time-lapses. Process is proof of human, and proof of human is the new trust signal.
- Develop a style a model can't average into existence. Weird, specific, personal. The middle is taken. The machine lives there now.
- Charge for judgement. Output is cheap now. Knowing what's good is not.
What this means if you're buying creative work
Artists aren't the only ones caught in this. If you run a business, you're on the other side of the same trade, and the repricing hits your brand too.
When every competitor can generate polished visuals for next to nothing, polished stops being a signal. Customers learn to read machine-made imagery as "ad" and scroll past. The work that stops them visibly came from a person with a point of view: a real photo of your real team, an illustration style nobody else has, a campaign idea that could only have come from someone who understands your customer. We make the case for that shift in the authenticity premium.
So when you hire creative help, change the questions you ask:
- Can they show their thinking as well as their outputs? Sketches, references, rejected directions. A creative who can explain why one option beats another is selling you judgement, the scarce part.
- Do they have a recognisable style? If their portfolio looks like a generic image feed, your brand will too.
- How do they use AI? "Never" is the wrong answer and so is "for everything". You want a clear line: AI for exploration and grunt work, humans for the decisions that define the look.
- Will the work still be yours in five years? A distinct visual identity compounds. A generated one can be matched by the next business that types a similar prompt.
Brands that invest in human-made, distinctive work now are buying the same thing the artists above are building: something that can't be averaged into existence. As we argue in why brand is the last moat, that's the one asset that gets more valuable as everything else gets easier to copy.
Art isn't dying. It's being repriced.
The camera didn't kill painting. It killed portrait-copying, and painters answered with impressionism. Recorded music didn't kill performance either, and live shows are now the main event. My bet is that AI does what those technologies did: destroys the market for the average and multiplies the reward for the exceptional.
Fewtrell is right. Art only dies when the artist dies. But careers die a lot easier than art does, and the careers that die in the next five years will belong to the people who let a text box do their seeing for them.
The pencil isn't obsolete. It just became the most contrarian tool in the building. Pick it up.
And if you want a brand built by humans who did the reps, with the taste to know exactly when the machines are useful and when they're poison, that's what we do all day. Our photo and video team shoots real people, real products and real places, and uses the machines only where they make the human work better.
Frequently asked questions
Will AI replace artists and illustrators?
AI will take over a lot of average, on-demand creative output, because generating decent images now costs next to nothing. It won't push out artists with trained taste, a distinctive style and the judgement to know what works and why. Those skills become more valuable as generic output floods the market. The risk falls on creatives whose only value was producing something passable quickly.
Is it still worth learning to draw in the age of AI?
Yes, arguably more than ever. Drawing trains your eye for composition, light and form, which is the judgement you need to direct AI tools well and to fix what they get wrong. If fewer people learn fundamentals, the ones who have them become scarcer and can charge more.
How should creatives use AI without losing their skills?
Treat it like an intern. Let it explore variations, mock up throwaway options and handle repetitive grunt work. Keep the decisions that define your work, such as style and the final call, in your own hands, and keep practising fundamentals so you can still judge and repair what the machine produces.
Is AI-generated art bad for a brand?
Not automatically, but it carries risks. Generated imagery tends toward a generic, familiar look that makes brands blend together, and audiences can learn to spot it and scroll past. Use AI for concepting and production tasks, and keep the visuals customers use to judge your brand, like your product shots and your identity, human-made and distinctive.