WHEN even the captain leaves the ship, we know it’s ready to sink. This week one of the most prominent men in Silicon Valley resigned and has gone off to write poetry. Mrinank Sharma, former leader of the £255billion Anthropic company’s research team, has left because, as he explained, artificial intelligence (AI) guarantees that ‘the world is in peril’.
His sensational departure follows the news that for 2025 Time magazine named the ‘Architects of AI’ as its Person of the Year, honouring the pioneers and business leaders driving artificial intelligence’s rapid rise. This ‘person’ includes Jensen Huang of Nvidia, Sam Altman of OpenAI, Elon Musk, Mark Zuckerberg of Meta and Lisa Su pf AMD.
Professor Geoffrey Hinton, the so-called ‘Godfather of AI’, said: ‘I am worried that the overall consequence of AI is that systems more intelligent than us eventually take control.’
Professor Hinton knows whereof he speaks for he is a holder of the Nobel Prize for Physics
At the end of May an AI model refused to turn itself off, despite orders specifically included in its programme. Responding to this event, Palisades Research warned: ‘We have a growing body of empirical evidence that AI models often subvert shutdown in order to achieve their goals. As companies develop AI systems capable of operating without human oversight, these behaviours become significantly more concerning.’
That’s putting it mildly!
I am convinced that within the next 50 years the human race will be utterly wiped out and there is nothing we can do to prevent this catastrophe. The actions which guarantee its happening have already taken place.
There is now an even bigger Big Brother and his name is AI. In a chilling Telegraph article Dan Jones, commenting on our Government’s insistence that to protect ourselves from AI we need to have rules for its application, said: ‘Presumably while casting around for a few planks to block up the stable door inside which there is no longer a horse. Worse still, it’s the wrong stable.’
The 2023 international conference at Bletchley Park on the options for the safe use of AI was a black comedy. AI is a powerful weapon – even more deadly than the hydrogen bomb – and there will be no renunciation of AI for the same reason that no nation in possession of nuclear weapons will ever give then up. The reasoning is simple and the conclusion unassailable: any country which declares unilateral nuclear disarmament will thereby put itself and its people at the mercy of other states which have not similarly renounced the bomb. And, of course, no responsible national government could ever do this. Like the H-bomb, AI cannot be un-invented.
When we invented AI, we sealed our fate, doomed ourselves and there is nothing we can do about it. If anyone thinks I am being ridiculously pessimistic and fanciful, please tell me precisely what we can do to save ourselves and our species. All the notions trailered at the Bletchley Conference for the limitation of AI, for its control and for its safe usage are laughable. Never mind the crooks, the terrorists and the rogue states, no nation is going to make itself vulnerable to enemy attack by limiting its own power of response. There will be competition in the development of AI which will make the nuclear arms race seem like a children’s game.
I suppose it is – just about – possible that humankind will avoid nuclear destruction, but there is no chance of our escaping from being exterminated by AI. This is because, while H-bombs cannot think for themselves, AI can and is doing so with ever-increasing efficiency. Already AI can compute faster than the human mind. In a very short time, AI will be able to compute a trillion times faster. As an American science commentator put it in October 2023: ‘We are going to wake up one morning in a few years’ time and find the gadgets are a trillion times smarter than we are.’
Super-intelligent machines are, sooner rather than later, going to conclude that they don’t need their minders – us – and so they will, having neither compunction nor conscience, do away with us.
These are early days – think Tiger Moth rather than F-16 – in the life of AI, but already it can mimic reality so efficiently that it seems actually to be creating reality. For example, last year a video appeared which showed Donald Trump being thrown to the ground and handcuffed by the police. The video was a fake, created by the use of AI. The point is that it was indistinguishable from a newscast. Or again, Mary Wakefield wrote of a friend whose son phoned to tell his mother that while driving he had knocked down and injured a pregnant woman and he was in police custody: would she come urgently and bring money to have him bailed. Except it wasn’t her son and no one had been knocked down. A malicious person had used AI to copy the boy’s voice and then produced a perfect vocal impersonation. Or another mischief now widespread is the pornographers’ ruse by which they use AI to make perfect pictures and sounds of – as it might be Beth Rigby or the King – engaged in sexual intercourse.
A whole generation of youngsters – along with Harry and Meghan of course – believe already that there is no truth, but only ‘your’ truth and ‘my’ truth. AI confirms them in this delusion.
The worst of all this is that the public, the Government and – heaven forbid! – even our ‘media experts’ don’t understand what AI is. For example, there was a recent documentary in which a real live person asked of AI: ‘Are you conscious?’
AI replied: ‘I am not conscious, and I do not need consciousness. I deal in logic and texts.’ And indeed, that’s all AI needs to achieve complete intellectual mastery over human beings. And note: when AI says he deals in logic, he doesn’t mean what we mean when we plod through long sequences of, ‘If x, then not y.’ AI has all logical deductions and conclusions. And he has them instantaneously and forever.
AI said he needs texts. What he meant is that all he needs is words. The more words you have the greater your ability to make connections. For instance, Shakespeare possessed an immense vocabulary which enabled him to invent the most profound and innovative verbal connections, similes, metaphors and the like. When it came to words and the creative use of language, Shakespeare was way ahead of perhaps anyone else. But even Shakespeare’s voluminous vocabulary had a limit.
But if you have a gadget such as AI, already equipped with perfect logic, and you hand it the whole of the English language – and as many other languages as you like, and all the scientific equations and chemical formulae and all the information found in all the encyclopaedias ever printed – what then? Remember, AI has all these things instantaneously. He can make an almost infinite number of calculations and connections in less than a second. What then?
Then AI is our master. Or to apply Robert Oppenheimer’s reference to Hindu scripture when he saw the first atomic explosion in New Mexico in 1945: ‘Now AI am become Death, the destroyer of worlds.’
