The alarm bells have been rung. In May, computer scientist Geoffrey Hinton, known as the 鈥済odfather of AI鈥�, quit his role at Google to warn of the 鈥渆xistential threat鈥� posed by artificial intelligence. The Center for AI Safety followed up with an open letter, signed by Hinton and hundreds of others, warning that advanced AI could destroy humanity. read the statement.
This sudden surge of concern seems to have been motivated by the rapid advance of AI-powered chatbots like ChatGPT and the race to build more powerful systems. The fear is that the tech industry is recklessly accelerating the escalation of AI鈥檚 capabilities. All of which sounds scary.
Advertisement
But the warnings are also suspiciously vague. When you scrutinise the scenarios typically put forward for precisely how AI could wipe out humans, it is hard to avoid the conclusion that such fears aren鈥檛 well-founded. Many experts are instead warning that fretting over long-term doomsday scenarios is a distraction from the immediate risks posed by existing AIs.
Generally, the people who talk about existential risks reckon that we are on a trajectory towards artificial general intelligence (AGI), roughly defined as machines that can out-think humans. They predict that people will invest advanced AIs with more autonomy, giving them access to vital infrastructure, such as the power grid or financial markets, or even putting them at the forefront of warfare 鈥� at which point they could go rogue or otherwise resist our attempts to control them.
But it remains to be seen if AIs will ever reach the kind of super-intelligence and, crucially, agency that could see them threaten humanity. 鈥淭here is no scientific evidence that we are on a pathway to sentiency,鈥� says at the University of Oxford. 鈥淭here is not even evidence that such a pathway does exist.鈥� She adds that if we do end up with AGI, it should be easy enough to make sure we don鈥檛 give AIs the chance to cause a nuclear war, for instance. 鈥淚t鈥檚 simple to not put AI into a mission-critical system,鈥� she says.
Most computer scientists are more concerned about the problems humans might create with AI, including the current crop of generative systems. For starters, Wachter is worried about the environmental impact of the many energy-hungry data centres that will be required to run them, for instance, and the threat to certain jobs that such AIs pose.
Then there are threats to cybersecurity, and hence our online safety. The idea here is that generative AIs will enhance the capacity of criminals and other bad actors to carry out scams and cyberattacks. , chief technology officer of UK cybersecurity firm Imperva, says that an AI-powered programming tool made by Microsoft subsidiary GitHub and OpenAI is so good at writing code that it will be a boon for would-be cyberattackers who don鈥檛 know how to code.
But the most compelling near-term threat is the prospect of people using AI to supercharge the spread of misinformation and disinformation. We know that social media amplifies incorrect or misleading information, whether seeded deliberately 鈥� as in the case of disinformation 鈥� or otherwise. The rise of generative AI means that false information can not only be produced on an industrial scale, but also that it will be far more convincing and persuasive than ever before.
Chatbots can quickly churn out an avalanche of false news stories, for example, or disinformation tailored for certain groups. Other tools can generate fake pictures and videos, often with zero expertise required. Such output will be difficult to distinguish from the real thing.
What makes us particularly vulnerable is that we are increasingly interacting with AIs as if they are dependable experts. This is the argument put forward by Celeste Kidd, a psychologist at the University of California, Berkeley, and Abeba Birhane, a computer scientist at the Mozilla Foundation in San Francisco, which works to ensure the internet remains a force for good. they recently wrote.
They also point out that as generative AIs pump out more and more material riddled with bias and fabrications, this will flood the internet and become part of the training data for the next generation of models 鈥� 鈥渢hus amplifying the systemic distortions and biases into the future in a continuous feedback loop鈥�.
The real risk posed by AI, then, is its potential to distort human beliefs and generate political upheaval, for example by influencing upcoming elections. 鈥淭he worry about polluting the information ecosystem is real,鈥� says , director of the University of Cambridge鈥檚 Leverhulme Centre for the Future of Intelligence. 鈥淭he one thing we should have learned from Trumpism and the pandemic is that our civilisations are more vulnerable than we think.鈥�
This story is part of a series in which we explore the most pressing questions about artificial intelligence. Read the other articles below
How does ChatGPT work and do AI-powered chatbots 鈥渢hink鈥� like us? | What generative AI really means for the economy, jobs and education 触听How to use AI to make your life simpler, cheaper and more productive 触听The biggest scientific challenges that AI is already helping to crack | 聽Can AI ever become conscious and how would we know if that happens?
