Like runaway climate change, the rapid development of self-learning artificial intelligence is an unprecedented existential threat to humanity, where past experience will be no guide to our future prospects (AI will end the west’s weak productivity and low growth. But who exactly will benefit?, 7 April). This is especially true when AI links to either super- or quantum-computing power.
Complex systems like these give rise to emergent properties, and circumstances where the whole becomes greater than the sum of its parts. Previously “dumb” neural networks like ChatGPT, by drawing on large language models, have already led to increasingly sophisticated and adaptable generative AI. As these systems become more complex and powerful, and their learning sources and human interactions multiply exponentially, it is reasonable to assume that AI may evolve its own consciousness and mind.
But it may not be one that we like. Society needs a moratorium on AI development, as called for by Elon Musk, Stephen Hawking and others, to decide what to do next.
Governments could, for example, move from taxing labour and work to taxing business AI, robot and software applications, especially those that displace human beings from the workforce. This would value human effort over machine contributions, and should help slow down the rollout of runaway AI as costs rise. The revenues could pay for an AI oversight agency, and retraining and other boosts to human wellbeing that Larry Elliott advocates.
Charles Secrett
Brighton
• Larry Elliott sees a future in which decision-making administrative tasks could be taken on by AI, thus putting thousands of white-collar jobs at risk. However, the most important lesson of lockdown was that human beings need other human beings, especially in classrooms, care homes, doctors’ surgeries and hospitals. It’s also hard to see how AI could fit into the equine industry or animal care, for instance.
So my rather polarising careers advice to students would be to either become higher-tech or more intensely human: to learn programming and coding to ensure they control the AI, or go into the most human and caring of callings, because that is where we will need the skills. This might therefore be a good time for the government to remodel its national workforce plans and ensure that remuneration is sufficient to keep doctors, nurses, animal care specialists and teachers happy in their jobs.
Yvonne Williams
Ryde, Isle of Wight
