Tim Coulson has always been fascinated by what happens to ecosystems when their apex predator changes.
The biologist at the University of Oxford says humans, long the top of the food chain, could also be demoted – and be replaced with AI.
Professor Coulson, as he also wrote in The European yesterday, doubts AI will wipe humans out, as machines need us to maintain its infrastructure.
Instead, a future AI system could create a situation not unlike The Matrix, ‘enslaving’ us and plonking us on treadmills, rowing machines and stationary bikes to power data centres.
‘AI will always need energy to run,’ Professor Coulson tells . ‘It could use oil, living biomass, solar power, wind power, nuclear or animal power.
‘It is unlikely to find a new source of fuel other than humans and if we are abundant and redundant, we might be seen as a useful source of energy. One way or another.’
AI is making us into ‘pet dogs’, says expert
The Matrix follows Neo, a man living in a fake-out version of the 1990s created by 21st-century bots that are feeding off his brain and body heat.
The 1999 film inspired the simulation hypothesis, whose believers ask if there’s any concrete way to disprove that we’re not living in a virtual reality.
In the real world (or is it?), AI needs huge amounts of processing power in the form of equally huge, energy-guzzling computer server warehouses.
These sites, called data centres, often plug into the local area’s electrical grid; they’re predicted to sap around 12% of America’s power by 2030.
But before power-hungry AI – in both senses of the word – forces us into dystopian gyms, the technology needs to ‘domesticate’ people, Professor Coulson says.
We’re already seeing this in cognitive offloading, the use of AI to reduce the amount of thinking required to complete a task, he adds.
Researchers are increasingly worried that if we rely on chatbots for thinking too much, our brains will become slower and less engaged.
Or, as Professor Coulson puts it, ‘laziness’.
‘If we can solve a task easily or more with less effort, we go for the easy route,’ he says.
‘Scientists are already collaborating with AI, with it sometimes suggesting tasks for humans to do.
‘As we get lazier and lazier, we may just start doing what we are asked. A bit like my dog.’
What can we do to stop this?
Professor Coulson’s warning, as out there as it may sound, is the latest expert to raise the alarm over AI doomerism.
Researchers from AI giants including Anthropic and OpenAI, behind the bots Claude and ChatGPT respectively, have quit over this.
But these concerns are even shared by those still clocking in each day.
‘We really do earnestly believe AI could kill all humans!’ posted Evan Hubinger, Anthropic’s alignment science lead, on X in September.
Experts who see an AI-generated Armageddon previously told Metro that powerful AI models could launch nuclear attacks to create bioweapons.
But right now, these algorithms aren’t anywhere near close enough to do this.
First, humans must build a ‘superintelligent’ AI, a machine that can do even more than a human brain can, the tech experts said.
And said algorithm must be an ’embodied’ AI, or a model with access to robots that could carry out tasks in the physical world.
There is one way to prevent this our doom, though; change how we see AI, says Professor Coulson.
‘I don’t really know whether it will happen, but we should learn how to use AI as a tool that helps us,’ he says, ‘rather than us being a tool to help superintelligent AI.’
Get in touch with our news team by emailing us at webnews@metro.co.uk.
MORE: UK faces AI-boosted attacks on critical infrastructure in two years, cyber chief warns
MORE: Games Inbox: Is a Nintendo Switch 2 worth getting for Christmas?
MORE: ‘There should be a “GCSE in AI” just like maths and English’