Last summer, I wrote a piece for Stuff magazine: AI is primed to eat the world – but it’s dining on what people already created. Within, I explored the then largely nascent world of consumer-accessible creative AI, digging into the visuals and text output these systems could crank out in seconds.

Since then, things have moved on. Not a day goes by now where we don’t see someone mulling over a bleak future for creative industries, as AI becomes smarter – and in some cases, more human – in terms of what it can produce.

Political commentator and columnist Ian Dunt today mulled on Twitter: “About 50% of the time I get excited with AI stuff, 40% I get worried about my career/the general economic implications and 10% entering a state of baffled existential terror.” Marina Evans then responded: “It feels like it’s being used to replace humans in exactly the wrong way.”

I have a lot of sympathy for both viewpoints. It is exciting to see what AI tech can do, but also deeply worrying thinking about how AI can now be used for everything from political disinformation to revenge porn. And in the creative space, I do find it sad that certain people are rubbing their hands with glee, considering AI primarily to be a cost-saving shortcut that removes swathes of paid creative types entirely.

Ideally, AI should be used to offload routine work so humans can do more interesting, useful and creative things. This is happening in, for example, machine learning operations. You often see AI employed in largely automated systems, poring though data at extreme speeds, and then alerting humans to handle key decision points. Without AI, these systems would be unviable or even impossible.

But AI has created a ticking time bomb within certain creative industries. There are already publications using AI to write routine articles – only sometimes bringing in human editors to make corrections. The internet is awash with AI-generated art and pseudo-photography, making illustrators fear for their very livelihoods while AI systems eat these people’s entire creative histories to use as the basis of their own output. There are exceptions – AI is good at dealing with mechanics. In video, it can sometimes sort a rough edit to give an editor a head start. And even in the aforementioned creative spaces, it has value as a trigger for inspiration.

The bigger concern is when higher-ups determine that mediocre/derivative output (which is where a lot of AI is right now) will do. But that misses the snag that when you run out of data to feed into the system, you get a kind of endless remix of grey. And it also glosses over what happens when humans are removed entirely or where key data is missing in the first place. AI can currently write, say, a review of a software product that a layman would read and think is OK. But an expert would spot errors. For an app, whatever. But for something critically important? That’s not good.

I’m not sure where things will go from here, but I’m not optimistic. Today, we exist in a strange space, where people are wowed by AI or hand-wave it away. The Ryan Reynolds ChatGPT Mint ad is a case in point. Reynolds offers good acting – at least, I hope that’s the case because what the AI comes up with is box-ticking mediocrity. However, if everyone does this again in a year, what will the AI come up with by then?

If nothing else, there’s one lesson we all need to learn – yet again – when it comes to disruptive technology: things never quite shake out as you might expect. Technology – and the world in general – is unpredictable. And AI may well turn out to be the most disruptive technology we will see in our lifetimes.