THE FACT OF THE MATTER
BY PAUL HOPKINS
The proposed EU legislation to limit access to social media, video-sharing platforms and online games for under-15s, a move Irish ministers have advocated for some time, is as clear as the proverbial mud.
How we will opt for a “gradual approach” with an effective ban on platforms has not been clearly thought-out and is open to manipulation, fake identities, skewed and toxic algorithms, parental onus and, predominantly, a lack of accountability from media giants like Meta and TikTok.
But worry not. For we are all doomed if Artificial Intelligence researcher Jacob Coxon is to be believed. Last month he quit US AI giant Anthropic, saying the firm and its competitors were “gambling with people’s lives”. His fellow researcher Evan Hubinger, in a post on X, declared: “We really do earnestly believe AI could kill all humans.”
Coxon isn’t the first to down tools over fears of AI doom and Man’s annihilation. The idea that AI could wipe out humanity was first advanced back in 2014 in Nick Bostrom’s book Superintelligence and the belief has long since circulated among AI scientists. The core idea is that AI, advancing at a rapid rate today, could in as little as five or 10 years become smarter than humans, escape our control and destroy our place in the scheme of things.
Earlier this year, Anthropic Safety Chief Mrinank Sharma departed, warning “the world is in peril”, while Alex Turner left Google DeepMind after the company signed a deal with the Pentagon permitting “killer drones”.
Coxon’s resignation has made waves, with more AI developers acknowledging the risk of “losing control and killing everyone”. They argue that AI extinction risk should rank alongside pandemics and nuclear war. Even Elon Musk has said the world needs to slow down and pull back on its “rapid” AI development.
Here lies the paradox: on the one hand, plenty of observers are scared by how AI is rapidly revolutionising industries, influencing productivity and shaping our future. On the other hand, AI has grown to be so ubiquitous that some people do not even notice its presence. We merge with our machines. Ultimately, they will extend who we are. Already, our mobile phones make us able to communicate with each other on a much better basis. It’s part of us. It might not be literally connected to you, but nobody leaves home without one.
To others, AI seems like a ‘black box’ in which mysterious things take place. And in the human mind, anything mysterious can easily trigger fear and distrust. In fact, for a significant number of people, AI has been transformed into some dark danger that’s lurking in the shadows
One of the biggest fears is, of course, job replacement. As machines perform tasks at lower costs and greater efficiency than was possible with humans, AI is already eliminating certain jobs and accelerating income inequality.
Other concerns tend to be more of an ethical nature. These fears revolve around a loss of control. Many worry that AI will create vast amounts of deepfake data, making it difficult to perceive what is real. As such, AI is already being easily used to orchestrate misinformation campaigns and cyberattacks. Those bots are very real – they are behind the scams you receive in your inbox.
There is the positive side, though. The scientific consensus is that, with AI, everything’s going to improve. Particularly in education and medicine, where remote surgery is now a possibility. We will be able to cure cancer and heart disease, and so on, using simulated biology – and extend our lives.
Meanwhile, the development of autonomous weapon systems sees AI research with all the hallmarks of an arms race – and the US doesn’t want to lose to China. The Trump administration shows little sign of slowing AI development at its most cutting-edge. Trump is trying to override state-level rules, arguing that caution risks losing to China.
History may offer us a template for how to manage a situation like this. Nuclear weapons are the classic case. Treaties and verification systems have not eliminated the risk of nuclear war, but they have slowed proliferation and no nuclear weapon has been used in conflict for 80 years.
AI capabilities have advanced faster than once predicted. ChatBot1 could barely put a sentence together – ChatBot4 can sit your uni exam. AI models will inevitably help build better AI models, and each generation will grow harder to monitor and control.
Without binding rules, we’re relying heavily on the goodwill of a handful of power-hungry companies.
