Right now, GenAI mostly produces cheap grey goo for a perpetual present

The main effect of GenAI right now is to reproduce stuff that is more or less indistinguishable from what already exists at no cost to businessesI keep hearing that Generative AI will unleash human creativity. Based on the contents of my inbox, I can tell you that is not yet the case. It is increasingly full of GenAI generated press releases and comment pieces that consist of a small amount of gibberish, a larger quantity of stuff that reads likes it’s been written by a non-native speaker, and an even larger amount of bland, clichéd C0nTeNt.   

I’m not going to blame the technology for this. It is people who are asking for it. And why wouldn’t they? Most of the stuff they’ve been producing themselves over many years has followed this template. Why pay a freelancer or PR to regurgitate some platitudes about how ‘hybrid working is here to stay’, when Chat GPT will do it for free in seconds?

The harsh reality for the people who have earned a living in this way is that if they think, talk and write like an automaton, an actual automaton will come for them sooner or later. Probably sooner.

It’s a point made powerfully by Rick Beato in this clip. He states that because people have increasingly made and consumed music that sounds like it’s been made by a computer, then eventually it will be. And is. … And that there will be an unfeasibly large amount of it.

 

Remember when everybody used to worry about nanobots turning the planet into a grey goo of organic matter? Instead of biomass, turns out we’ve been creating it for ourselves in the form of content, entertainment and art, and now we’re asking machines to produce it for us in vastly increased quantities.

The effects are already apparent. The offer of work on digital freelance platforms is already falling away, not that they were ever known for either the supply of – or demand for – original thought and creativity.

The main effect of Gen AI right now is to reproduce stuff that is more or less indistinguishable from what already exists at little or no cost to businesses

The upshot is that the main effect of GenAI right now is to reproduce stuff that is more or less indistinguishable from what already exists at little or no cost to businesses. It’s a point made powerfully by Ted Gioia here. And is apparent in the behaviour of Adobe who have been offering images in the style of the photographer Ansel Adams until shamed into backing off. Image libraries are increasingly packed with AI generated images.

All of this has a long way to play out. No doubt some people will still value authenticity and seek it out, perhaps by turning to nostalgia. This might include for a world that predates them, as is the case with my own 14-year-old daughter whose musical and fashion tastes are rooted in the 1980s. It is also apparent in the growing demand for ‘old music’.

When it comes to the written word, we face other problems, notably the fact that there may be no new material with which to train AI very soon. A group of researchers has published a paper which says that high-quality language data that can be used to train GenAI will be exhausted by 2026.

So something will need to change if we are to see the supposed effects on creativity. For now, the promise of AI is to trap us in a perpetual present, and we may already be there.

Main image: The Torture of Prometheus by Gioacchino Assereto