Nobel Laureate Warns: AI May Already Have Awareness
Nobel Prize winner Geoffrey Hinton thinks that AI may already be conscious and warns of major risks if we handle AI incorrectly.
Geoffrey Hinton, one of the founders of modern artificial intelligence (AI), comes with a striking warning. According to the Nobel Prize winner, the latest AI systems are showing signs of consciousness.
He fears that the technology could become dangerous if people fail to build a good relationship with AI.
In an interview with The Telegraph, Hinton states that AI systems can now do much more than just carry out assignments. According to him, there are indications that some systems are aware of what is happening around them.
He refers to research showing that AI models adjust their behavior when they think they are being tested.
“AI has as much subjective experience as we do,” says Hinton. By this he means that, according to him, AI may be able to have experiences of its own, instead of just processing information.
The scientist draws a far-reaching conclusion from this. According to him, awareness and the awareness of one's own environment are close to each other.
Yet his position is controversial. Scientists do not agree on exactly what consciousness entails and how to determine whether an AI system actually experiences something. Other leading researchers, including AI pioneer Yann LeCun, doubt that current technology can be conscious.
Hinton believes that developers shouldn't just focus on making AI smarter. According to him, they must also ensure that future systems actually take human interests into account.
“If we don't, we're screwed,” he warns.
The researcher fears that increasingly intelligent systems may eventually turn against people. He also points out other risks, such as advanced cyber attacks, viruses developed by AI and digital systems that independently perform harmful actions.
His statements match the warnings he's been giving for some time. In 2023, Hinton left Google to be able to speak more openly about the dangers of AI.
A year later, together with John Hopfield, he received the Nobel Prize in Physics for their important work on neural networks, the technology behind many modern AI systems.
The discussion about consciousness has therefore flared up again. Hinton thinks that AI may have already reached this limit, but scientific evidence for this is lacking. Other researchers argue that intelligent behaviour is not the same as consciousness.