AI has captured the imagination but can barely replicate one
I’ve been wondering how best to keep ahead of AI. That doesn’t seem strange to you, does it? Don’t get me wrong, I understand why some people think it’s wonderful to be able to type something into ChatGPT and get a rapid readymade answer or explanation in response. It’s kind of like Wikipedia except you don’t even have to go to the website to find anything out. How convenient.
By now, we’re all familiar with some of the issues plaguing AI: its tendency to hallucinate and to provide seemingly authoritative statements which, to the unwary, might be treated uncritically as cold hard fact when they are anything but. The widespread dissemination of AI-generated material, whether it’s written, spoken or visual, is threatening to create a crisis of reality when our ability to tell truth from fiction is being undermined on so many fronts.
It wasn’t supposed to be like this. We can all agree on that. In fact, if you listen to the tech companies ramming AI down our collective throats, it still isn’t. The problem is that, to all intents and purposes, AI is being misapplied. The people hyping AI hype are focusing on making it run when they should be starting with trying to make it walk, preferably in a straight line.
I don’t blame them, necessarily. They want to make it exciting, to capture our collective imagination. To a certain extent, they’ve succeeded but in a very different way.
Lots of us expected AI to be focused on the grunt work. The tasks that most of us don’t want to do and that we don’t want anyone else to have to do. God knows, there are plenty of those tasks out there.
Harvesting v theft
But the people selling AI had a very different idea. They wanted it to, literally, capture our imagination. First, by harvesting (okay, stealing) as much of the creative output of humanity they could get their hands on to ‘train’ their generative AI. While there is growing criticism of this theft, the companies behind generative AI are still trying to rewrite the copyright rules to continue their wholesale pillaging of human creativity. They want us all to keep feeding the plagiarism machine and too many politicians are bending over backwards to help them.
So far, no one has managed to explain just why artists and creatives should agree to this state of affairs, especially as the result is likely, at least for the interim, to be flashy mimicry of their creativity on a mass scale.
Which brings us to the other quite literal means by which AI can capture our imagination. Artistic mimicry, especially uncredited or by means of deception, creates a scenario where the line between real and artificial becomes blurred to the point of meaninglessness.
We have already seen the consequence of machine learning on a large chunk of the written content we read on a daily basis. Consider how much of what you see on the web is constructed to aspire to score higher against SEO criteria rather than written to be the most engaging and well-written that it can be. The result is a deluge of formulaic writing that might as well be constructed by a machine.
Now, that trend is moving to music and the visual arts too. To focus on music, for example, the streaming platforms have played a major role in influencing the development of many of the songs we hear. They have incredibly short lead ins, vocals frequently employ autotune, instruments are often digital midi representations rather than human players and they follow entirely predictable formulas.
That’s not all the fault of machines, the fact is people like formulas. That’s why so much of our entertainment follows a small set of formulas. That’s why there are supposed to be only seven plots. But there is still plenty of scope for creativity.
Or there should be. Unless the space becomes crowded with AI-generated content. And that looks like a very real danger judging by reports from Deezer that more than 18% of all new music content uploaded was fully AI-generated, representing around 20,000 tracks a day. Flooding the zone with fake music threatens to crowd out some of the people trying to create their own music and prevent them from getting noticed or played. The rapid escalation in this AI-generated music – in Deezer’s case it went up from 10% in January to 18% – does not augur well.
How do we escape this onslaught? By being different? Avoiding cliche, perhaps? Inserting a degree of personality into what’s being created that AI can’t replicate? Will the people reading it or listening to it be able or willing to tell the difference? If they’re not, whose fault is that?
Where is there left for us to go if we let AI capture our imagination? Let’s all try to keep ahead of AI as best we can.
Subscribers 0
Fans 0
Followers 0
Followers