AI was supposed to be the shiny centrepiece of the new American tech revolution - until it wasn’t.
This week US tech giants took an existential hammering, seeing a trillion dollars wiped off their market value. The absurdly valuable Nvidia lost $600 billion all on its own. The AI sector is now in a state of freefall.
OpenAI and its ChatGPT model have had hundreds of millions of dollars invested in its development. It can begin to feel like meaningless large numbers attached to these things. It might as well be.
Because China has made a bombastic entry in the race with DeepSeek, an efficient free-use generative AI model with a paltry supposed development of $6 million. The bloated and overvalued US tech sector is tumbling from this unexpected lean visitor.
Read more:
This film veteran was stunned when he asked AI to give him ideas
So curiously, I asked the market-wrecking DeepSeek to write an opinion column about AI in the style of Derek McArthur from The Herald. This column has plenty of text data for it to feed from by this point, and I’ve written about AI and its effects on arts and culture multiple times in the past, so it should get pretty close to the real thing, right?
The results were interesting, but not in what it got right about my thoughts on AI and the way I’d express them, but in what was missing. There was something off.
The article DeepSeek spat out at me had largely the correct perspective and picked up on thematic threads built up over time – it even used words I’ve added to a ‘do not overuse’ list.
“One thing is clear: the rise of artificial intelligence is not just a technological phenomenon, but a profoundly human one”, writes the AI. “It is a mirror reflecting our hopes, our fears, and our values. Whether it becomes a force for good or ill depends not on the machines themselves, but on us.
“As we harness the power of AI, let us ensure that it is guided not by cold logic alone, but by the warmth of our humanity. In the end, it is not the technology that defines us, but how we choose to use it.”
It reads like my writing voice, but it certainly doesn’t feel like I wrote it. There is an appeal to humanity in its argument but there remains a soullessness. It’s clear we are not dealing with sentient thinking quite yet.
It's been claimed that development for DeepSeek only cost $6 million (Image: Reuters) AI is still missing a full comprehension of how humans think, the areas of thought that can’t simply be reduced to data and regurgitated in new forms. Whether that’s even possible is an overarching question for AI’s future, but then no one expected China to suddenly deliver a model with such low-cost efficiency.
The space where thoughts and ideas can be fluid still eludes these models, where nuance can be introduced, considered, and broken down. The writing style and the sentiment are correct, but the way it’s expressed leaves an uneasy feeling of uncanny valley. It attempts to replicate a sense of my humanity and tie it to the subject of AI, but it fails.
“As a nation with a proud tradition of innovation and a burgeoning tech sector, we have a unique opportunity – and responsibility – to shape the future of AI in a way that reflects our values. The Scottish Government has already taken steps in this direction, with initiatives aimed at fostering ethical AI development and ensuring that the benefits of this technology are shared equitably. But more must be done.”
Again, the correct perspective, but it doesn’t realise how that perspective fits into an overall branch of thinking. True, I think AI is a tool that can be dangerous in the wrong hands and useful in the right ones, but it lacks the ability to be sufficiently self-critical. It can critique itself, but only in the way of talking points.
And I certainly don’t think the Scottish Government has any sway in changing or influencing the ethical boundaries of AI even if it has found that line of thinking from thematic threads of the past. Unless this column is just the same points repeated in the same contexts over and over, it will struggle to represent arguments correctly and put them in their proper places.
Read more:
Derek McArthur: OpenAI’s Sora could kill commercial film – good riddance?
The same problems with ChatGPT are also apparent with DeepSeek. Despite the scraping of my words, phrasings, and perspectives, it still reads much like a barebones university essay, complete with “in our brave new world” cliches. Truly a land of contrasts.
AI has infected the arts in a largely negative way so far – streaming services like Spotify are filled with AI-generated music for the purpose of receiving pennies in royalties, AI elements hastily incorporated into films has caused controversy and disappointment (with the latest being Best Picture nominee The Brutalist), undiscerning art shops have AI art adorning their selection more frequently now, and literature sections of sites like Amazon are littered with AI-written books as part of a strange online course hustle. Oh, the grifters can make a quick buck? Art be damned then.
Luckily, it seems the robot replacement for this column can still be kept in the storage cupboard collecting dust. But there remains a lot of space for these types of models to be refined, expanded, and gain further sophistication. It would be unwise to underestimate what's coming next.
Share