There is lots of discussion on the use of AI and its impact on creative industries. Authors, artists and musicians are voicing their concerns particularly loudly, and have celebrity support from Dua Lipa and Elton John. One issue surrounds the legal implications of model training on copyrighted material, and another is a more general worry about AI-generated content replacing real human work.

The copyright question will have to be answered by the courts and legislators, ideally with global consensus, and by balancing perceived fairness with the pragmatic realities of hugely useful new technology. The focus should be on outputs more than AI training, as this aligns much more closely with existing concepts. No one stops you reading an article, learning from it, and being inspired to write your own. You only end up in hot water if “your own” strays too close to the original, and I don’t see why similar thresholds shouldn’t apply to AI.

Public debate around copyright has also failed to recognise the nuance and breadth of AI, a term that captures a wide range of technology and applications. Models that generate gimmicky images are not the same as models that are trained to analyse medical scans, but both require large amounts of data. Legislation that tars everything with the same brush is unlikely to be optimal.

Creatives fear that they could lose their jobs, replaced by the very robots who stole their valuable content in the first place. This concern and a sense of deep injustice is what has driven many to dive into the copyright debate, instinctively seeking to block the threat at its source.

History is littered with examples of technological advancements that turned an existing industry on its head, changed the world, and ultimately made it a better place. These historical analogies have many interesting parallels with the AI story, but also highlight important differences.

19^th century artist Paul Delaroche is said to have exclaimed “”From today, painting is dead” when confronted with an early photograph. Fortunately, we didn’t ban the camera to protect him. Almost 200 years later painting still happens, people still buy art, and a whole new industry, a new photographic art-form, now exists.

From cutlery to clothes, manufacturing efficiencies have put many handcrafters out of business over the years. It’s a familiar pattern of initial shock, disruption, and ultimately adaptation.

Innovation forces consumers to judge value harshly. Is this portrait expensive because it’s a beautiful piece of art, or is it expensive because it took 2 months to paint? Is this knife expensive because it’s a stunning piece of craftsmanship, or is it expensive because they can only make 10 in a year? The average local portrait painter bought a camera, and those who had hand-made knives that didn’t pass the test moved onto the factory floor.

In all these cases, the incumbent craftspeople and creatives didn’t disappear altogether. Artists whose work was deemed to have a magical quality that a photo couldn’t replicate stuck around, and probably enjoyed elevated status as a result. Truly superior handmade products still command a premium over mass produced alternatives.

It’s logical to assume that AI is the latest technological leap, perhaps unusual in the breadth of its applications, but otherwise similar to our examples from history. Creatives who fail the knife test may drop away, while those who pass can thrive as artisans. Concerns around how difficult it can be to distinguish between AI and human content miss the point; if your work is no better than AI’s it’s a sure sign that you fall on the wrong side of the divide.

It’s not all bad news for those professional creators who know they will be in the group who lose out. As the value of mediocre creative output falls through the floor, there will still be a place for those who embrace AI tools to radically increase their output. The Jevons Paradox means total demand for writing, images and music will increase, albeit now served by a smaller number of human professionals.

Real human art, in all its forms, is about more than just the end result. If it wasn’t, the art market wouldn’t be so irrational. Human creative expression will always have an intangible quality and will remain strongly in demand. Professional creatives will also be aided by the fact that commercialisation relies on a degree of in-person distribution. Musicians play live gigs, authors sign books, and artists sell real, physical paintings. By contrast, a faceless composer dreaming up mundane music for the background of car commercials might well find themselves replaced by, or using, AI.

I would go out on a limb and say that most creative expression comes from hobbyists, those doing it not because they expect to make any money, but because they want to. People paint, blog, and learn guitar for fun. If anything, AI should help these pursuits by providing easier access to advice and inspiration. The passionate amateurs may even find themselves at a growing professional advantage as a result of their continued creativity.

The risk of model collapse is a unique AI phenomenon that will also serve in old-fashioned creativity’s favour. In the longer term, AI’s need for high quality human training data will act as a natural counterbalance to its domination, much like other ecosystems that support just the right proportions of a diverse food chain. Like any human intervention in nature, the answer to the copyright legislation question will have an artificial impact on exactly where the balance lies.

If the top professional human creators will enjoy elevated status, the hobbyists will be free to continue, and the mediocre are freed from their purgatory, then the creative industries have little to fear. Those in white-collar jobs where there has never been much value in human expression are the ones who should be looking over their shoulders.