More than half of listeners say they care if music is AI-made. Deezer found 97% cannot tell

The BBC published an InDepth essay on Sunday 11 October by Ros Atkins, presenter of The Value Crisis, asking whether AI is creating a "human premium" for art made by people. His opening numbers come from the music platform Deezer. It says 44% of tracks uploaded to it are AI-generated, and its research found that 97% of listeners could not tell which tracks were AI-generated. More than half of those surveyed said they did care whether a track was made by AI. They want humans involved.

The gap between those two numbers is the story. People say they value the human hand. Their ears cannot find it.

The premium attaches to the process, and only where the process is the product

Daniel Susskind, an economics professor at King's College London, separates art from more functional, practical tasks, a distinction that matters for anyone outside the arts. He recalls an accountant telling a room that clients came for "the personal touch, the human process". Susskind told him the clients came because they wanted their taxes done properly.

Art is different, Susskind argues. Under the Sistine Chapel ceiling, part of the awe is knowing that a human mind dreamt it up. He points to chess. Garry Kasparov lost to Deep Blue in 1997, and today millions play the game, because people care that a human being is moving the pieces.

Read that as an operator. The human premium is real, and it is narrow. It pays where the buyer values the person doing the work. In functional work, such as the tax return, the report or the reconciliation, the buyer values the outcome, and AI can produce that outcome at scale. Selling "the personal touch" on a deliverable the client judges on accuracy repeats the accountant's mistake.

Kyle Chayka of The New Yorker describes where the premium is forming. He says AI writing is now widely considered "generic, obsequious and banal", so readers are coming to value obviously personal writing, even when it is messier. He says designers are leaning into work that is "weirder and more eye-catching" to find what AI cannot generate easily. Labels like "AI-free" already appear on the covers of some books, films and albums. Alex Mahon, chief executive of Superstruct Entertainment, has noticed more appetite for live, unpredictable experiences.

The algorithm already took the curation layer

Atkins doubts the premium will hold. AI content is growing so fast that people may soon struggle to tell what involved AI, and the way people find culture has already changed. People used to seek culture out. Increasingly it arrives in an endless stream, sorted by an algorithm that predicts their taste. Chayka, author of Filterworld, says algorithms have made us lazy and "prevent us from going deeper". Creators then shape their work to chase the attention the algorithm hands out.

That loop is the bigger threat to the premium. More than half of Deezer's listeners say they care whether a track is AI-made. If 97% of listeners cannot tell the difference, and recommenders do more and more of the discovering, that preference has little way to act. A preference buyers cannot check is hard to charge for.

The author John Lanchester makes the hopeful case: when one kind of thing gets commodified, the things that cannot be commodified start to seem more important. The BBC's evidence supports both readings. Your job is to know which one applies to your work.

Your Next Move

• Sort your output into two piles. Pile one is work the buyer judges on the result. Pile two is work the buyer values because you did it: the judgement call, the relationship, the room you are in. Let AI do the routine work in pile one and spend the time you get back on pile two. The step-by-step method for running that sort on your own role is in our afternoon audit of which tasks in your job AI is absorbing.
• Make the human part visible. Deezer's listeners cared and still could not tell. If your contribution disappears inside the finished work, put your name and your reasoning on it.
• Protect your own taste. Each week, take one recommendation from a person you trust instead of a feed. Judgement is built from inputs you chose yourself.