AI is taking up more and more space in our society. It only took a short time for technology to invade every sphere, causing concern. Notably that of Geoffrey Hinton, winner of the 2018 Turing Prize and the 2024 Nobel Prize in Physics and pioneer in the field.
- Geoffrey Hinton warns of a risk of extinction of humanity by AI with a probability that he estimates between 10 and 20% in the next thirty years
- AI pioneer criticizes lack of strict international regulation and calls for more safety research to be imposed on companies like OpenAI
- Some experts like Yann LeCun believe that AI could on the contrary save humanity from extinction, an opinion which diverges from the fears of Geoffrey Hinton
Geoffrey Hinton is not his first worrying prediction about AI. During an interview with BBC Radio 4the expert estimates that there is between 10 and 20% probability that AI will lead to the extinction of humanity in the next thirty years. Previously, his prediction was 10%.
Why is AI threatening humanity with extinction?
In October 2023, Geoffrey Hinton warned about X, explaining that poorly regulated AI presented a one in ten chance of destroying humanity within three decades. According to one of the sponsors of AI, regulation is a crucial issue since it is an unprecedented situation in history: never before has humanity had to control an entity more intelligent than itself.
Geoffrey Hinton explains that when faced with a three-year-old child, “we are in control”, but when faced with an AI, “we are the child of three”. The scientist is particularly concerned of the race for the benefit of large companies (like OpenAI) which could lead to the creation of uncontrollable AI. The solution: impose more safety research on these companies through strict government regulation.
Currently, only Europe has adopted binding legislation with the AI Act. The lack of international regulation leaves companies self-regulate according to their own ethical criteria. This is an unprecedented situation which is causing some commotion. For example, Grok has few limitations and can generate misleading images of real public figures. What a Dall-E refuses to do, for example.
AI, the most dangerous invention in the history of humanity?
Geoffrey Hinton’s position is not new, far from it. Yoshua Bengio, another major figure in AI, believes that this technology is the most dangerous invention in human history. Before the nuclear bomb? It seems so, according to concerned experts.
A global conference on AI will be held in Paris in February 2024. This could be a turning point since many heads of state will be present. On the path to unified international regulations to address the concerns expressed by technology pioneers?
Note that other experts are more optimistic. Yann LeCun, also considered one of the technology pioneers and chief AI scientist at Meta, believes that the technology could save us from extinction. Quite the opposite of his colleagues.
