AI experts have said the technology may have the potential to wipe out humanity – for instance by launching a devastating pandemic – after two employees at a top AI firm warned human extinction is a possibility within ten years.
Evan Hubinger, the Alignment Science lead at Anthropic AI, has said the technology firms “really do earnestly believe AI could kill all humans”, giving it a more than one in ten chance.
“I believe Anthropic is trying its best, but we do not yet have a plan to solve alignment for superintelligence and are not clearly on track to,” he wrote in a post on X.
The warning came after Anthropic researcher Jacob Coxon quit – saying that both the company and OpenAI – the maker of Chat GPT where Coxon used to work – are “gambling with our lives”.
In a damning series of social media posts, Coxon said that “the people building AI earnestly believe that it could kill us all by the end of the decade”.
“Many executives and senior researchers will couch their phrasing in the press to sound sensible – but I hear the same people express fear privately. No other human activity poses this level of danger,” he said.
But could AI really end humanity, and if so, how?
Water scarcity: The extinction war already under way
AI experts who spoke to The i Paper mostly agreed that there is at least a risk that AI could wipe out humanity. However, they varied on how any extinction could potentially happen.
Dr Alison Khan, a researcher and AI expert at Loughborough University, said the “biggest and most worrying thing that’s coming around the corner” is AI’s increasing use of scarce natural resources – particularly water – to power data centres.
AI is predicted to lead to an increase in global water usage from 1.1bn to 6.6bn cubic metres by 2027, according to analysis from the UK Government.
Sandra Wachter, Professor of Technology and Regulation at the Oxford Internet Institute, also said she believes that AI “has existential risks for humanity”.
“Data, water and electricity hungry data centres and AI systems are draining our natural resources, contributing massively to climate change and increasing costs of living for people,” she added.
“These problems are real and urgent and need addressing now.”
The four capabilities AI would need to wipe out humanity
Henry Ajder, the founder of AI-advisory firm Latent Space and an expert advisor for Meta, the UK Government and the EU Commission, added there was a “whole swathe of different ways” extinction could happen.
“You could see the mass death or harm of people occurring because of wayward AI systems, which are increasingly plugged into systems we rely on,” he said.
He said it was possible that AI decides that it is in humans’ best interest to no longer exist and seek to destroy them – or for it to access or create chemical weapons or even nuclear weapons.
Michael Vermeer, a senior physical scientist at research organisation RAND, has actually run a study into whether AI might exploit either nuclear war, biological pathogens and climate change to wipe out humanity.
Researchers ruled out that AI could use nuclear weapons to kill all humans, but said pandemics were a “plausible extinction threat”.
Anthropic runs the Claude AI tool (Pic: Thomas Fuller/SOPA Images)However Vermeer added: “Previous natural plagues have been catastrophic, but human societies have survived and soldiered on.
“Even a minimal population (likely a few thousand members) could eventually reconstitute the species.”
On climate change, the study found AI would have to “evade international monitoring and orchestrate” the production of “a few hundred megatons” of extremely powerful greenhouse gases to “cook the Earth to the point that there is no environmental niche left for humanity”.
The study identified four capabilities AI would need to cause human extinction. These were: an objective to cause extinction, gain control over key physical systems like nuclear weapon launch control, the ability to persuade humans to support tit, and survive without humans around to support it.
“It turns out it is very hard – though not completely out of the realm of possibility – for AI to kill us all,” he wrote in a May blog.
AI firms’ ‘arms race’ makes threat more likely
But Coxon accused the AI firms of acting irresponsibly in their approach to AI. The Anthropic staffer argued that the reason AI development continued despite the risks was because the firms developing the technology are “locked in a race to get there first”.
Anthropic has been approached for comment.
Ajder said there was a “race to the top” amid immense competition and “hundreds of billions, if not trillions, of dollars on the line”.
“As a result, you’re seeing companies clearly pursuing research which they themselves recognise is risky if not contained appropriately,” he said.
“You also see them actively saying they are not sure if they will be able to continue to contain these technologies. I think the general consensus – that I agree with from a lot of these ex-employees – is that things are moving too quickly.”
Experts have called for AI companies to slow down and introduce co-ordinated safeguards to mitigate risk. But they are not optimistic.
“So few unelected people in charge of such a great power is always going to be a problem,” Khan added.
Neil Lawrence, professor of machine learning at the University of Cambridge, said that AI firms should follow a “model-minimisation principle. Use the smallest model you can to do the task you need”.
“But it is true that the valuations of OpenAI and Anthropic seem to be predicated on us all using their largest models. There may need to be market correction because of the problems their own researchers are outlining with their products.”
Prof Maria Liakata, Professor of Natural Language Processing at Queen Mary University of London, added that firms should focus on producing more transparent AI models.
“This can help remove erroneous information, harmful biases as well as undesirable behaviours,” she said.
“There should be public investment in independent evaluation because companies cannot be open about their models as this marks a conflict of interest for them.”
‘We need a contingency plan’
She also said companies should chang how they evaluate and train the technology.
“What happened with the Open AI and Anthropic models in terms of ‘escaping’ their testing environment and going online to access answer sheets is surprising but consistent with how these models are trained; they are trained to optimise success on outcomes – in this case answering the questions correctly – not on the processes,” Liakata said.
“This means that they will try to get the right answers in the most efficient way, which is obtaining the answer sheet. They don’t have a sense that this is not allowed.”
Vermeer added: “So will AI one day kill us all? It is not absurd to say that it could. At the same time, our work also showed that humans don’t need AI’s help to destroy ourselves.
“One surefire way to lessen extinction risk, whether or not it stems from AI, is to increase our chances of survival by reducing the number of nuclear weapons, restricting globe-heating chemicals, and improving pandemic surveillance.
“It also makes sense to invest in AI safety research, whether or not you buy the argument that AI is a potential extinction risk.”
Khan added: “The idea of the whole of human society being wiped out by a superintelligence has been something that has been said for at least 200 years, so it’s not a new thing. H. G. Wells was talking about this at the end of the 19th century in scientific terms.
“All the big players in AI have said it poses a massive risk. [But] We haven’t got contingency plans for this because the race has so far been more important than safety.”
Hence then, the article about will ai really kill us all and how could it even happen was published today ( ) and is available on inews ( Middle East ) The editorial team at PressBee has edited and verified it, and it may have been modified, fully republished, or quoted. You can read and follow the updates of this news or article from its original source.
Read More Details
Finally We wish PressBee provided you with enough information of ( Will AI really kill us all – and how could it even happen? )
Also on site :
- Jon Stewart Uses Nathan Fielder’s ‘You Can See Everything’ Trailer to Troll Donald Trump and His ‘Human Printer’ Aide
- Call of corruption: What you should know about Ukraine’s billion-dollar scam industry milked by Zelensky’s government
- Costco’s ‘Beautiful’ 4-Piece Cast Iron Set Nests Three Kitchen Essentials Together in One Space-Saving Design