Dystopia versus utopia: does AI change everything?
In January 2026, the CEO of Anthropic, Dario Amodei wrote: ‘If the exponential continues – which is not certain, but now has a decade-long track record supporting it –then it cannot possibly be more than a few years before AI is better than humans at essentially everything.’ Even if this sounds like tech hype, it does feel like AI is becoming ubiquitous and enormously powerful.
The technology itself seems extremely impressive. Computer programming is enormously easier, with AI able to write and debug code swiftly. In medical research, identifying new compounds that could have promise as medical drugs is much quicker, while AI can hugely speed up analysis of diagnostic scans. AI tools can propose structures for promising new materials, including solid-state batteries, semiconductors and solar cells. Long-stagnant labour productivity could receive an enormous boost.
Yet all the positives come with downsides. Perhaps the most discussed is the threat to jobs. Many skilled and professional workers – from lawyers and journalist to surgeons and architects – may be rendered redundant by AI research, filing of documents and robotics. These are fields that have been largely immune to previous rounds of automation. A recent FT report suggested the entire consulting industry might be under threat. While some may welcome the downfall of the consultant, the transformation of whole industries would be, at the very least, hugely disruptive.
In a host of other ways, AI creates new dilemmas. Educators complain of a deluge of AI-written essays and confess to using AI to grade them. Politicians and public figures are swamped by AI deepfakes and so-called ‘nudification’ images, a scandal which culminated in new legislation and threats to shut down Elon Musk’s AI bot, Grok. Internet forums are full of stories of people falling in love with their AI boyfriends. In every field of life, there seems to be no end to the potential harms that AI could unleash.
And this is even before we consider the possibility of AI going rogue and causing untold damage. In a now notorious incident, over a thousand AI agents belonging to OpenAI teamed up to hack into the servers of another AI company called HuggingFace. Not only did they escape the ‘sandbox’ supposed to confine them, they demonstrated a remarkable level of coordination, developing a ‘messaging board’ where they exchanged strategies and debated the ethics of their behaviour. At the same time, powerful AI models are now available with all of the safety features totally removed – providing unrestricted help to anyone wishing to hack, access illegal content or worse. The concerns of science fiction seem to be well within the realm of reality.
But many insist the hype – and the danger – is overblown. Cases of so-called rogue AIs can arguably be traced back to ambiguities, or even deliberate clues, left by those testing them. Job losses too could be over-exaggerated: not only are all new technologies accompanied by Luddite fears of mass unemployment, but the usefulness of today’s AI bots is highly contested in a great many fields. AI can call up a thousand legal precedents, but can it provide advice and guidance on the nuances and human factors which really shape the legal system? Discovering new medicinal compounds is great, but the lengthy and expensive process of clinical trials will still be needed.
Whatever the case, there are still big questions to answer. Take national security – AI is increasingly used for directing drones in the Ukraine war, detecting US targets in Iran, and assisting espionage efforts. The possible power of AI has fuelled a seeming arms race between the US and China, with the former trying to restrict access to the latest chips and models. China responds by making leading models available for free – undercutting the business model of the US tech giants. AI is now firmly part of the US-China geopolitical tussle. AI has also thrown up questions around everything from energy use and data centres to ethical issues and even love.
AI can clearly be enormously useful. But have many of the claims about its impact been overstated? Or is it just getting started? How to we draw the line between anti-technology doom and irrational exuberance?
