“10 to 20 % chance that AI eliminates humans”: the sponsor of the AI delivers his method to avoid the worst

Geoffrey Hinton is nicknamed the “AI godfather” because he found ways so that computers learn by themselves, a bit like a human brain. Since major advances such as the arrival of Chatgpthe is regularly questioned on the subject. Like many, it is aware of the enormous importance that artificial intelligence has taken in our lives and the sometimes disastrous consequences that it can have. However, he has a solution to survive.

Chatgpt

Chatgpt is the Openai chatbot, based on the GPT artificial intelligence model, making it possible to answer all kinds of questions or requests. Available in free online version.

• License:
  Free license
• Author :
  OPENAI
• Operating systems:
  Windows 10/11, Apple Silicon macOS, online service, Android, iOS iPhone / iPad
• Category :
  IA

Geoffrey Hintonoften presented as “sponsor of artificial intelligence“, is worried about the dangers that technology could represent and in particular artificial intelligence. Winner of the Turing Prize and former researcher at Google, he estimates that it exists between”*10 and 20 % like that the‘IA eliminates humans*”If no effective measure is taken.

For him, the efforts of the tech giants aimed at keeping the machines “under control” are doomed to failure. “” “It won’t work. They will be much clearer than us. They will have all kinds of ways to get around this”He warned at the AI4 conference in Las Vegas.

Fortunately Hinton has an idea: to program AI systems with “maternal instincts”The goal would be that they develop a real attachment to humans, even in the event of intellectual superiority; a bit like a mother and her child. According to him, an AI conceived to protect and take care of its users would have fewer reasons to want to eliminate them.

#240903 – News | Jealous of Chatgpt, Elon Musk threatens Apple with a trial that could cost him dearly.

An opinion that is not unanimous

However, this vision does not convince everyone. Fei-fei li, nicknamed the ” IA godmother“, Rather defends the idea of an AI” centered on human “, which guarantees the dignity and autonomy of people.

For its part, Emmett Shear, former temporary CEO ofOPENAIrecalls that some models of AI have already shown their ability to deceive or manipulate humans, in particular by lying or threatening them.

For Hinton, the threat is closer than you think. While he previously believed that general artificial intelligence would not arrive before 30 to 50 years. Today we are talking about five to twenty years old.

He nevertheless recognizes the positive potential of this technology, especially in the medical field, where it could allow the creation of treatments against diseases like cancer. But its warning remains the same. If we do not find a way to make AI fundamentally benevolent, it could judge that we are no longer necessary.

Want to save even more? Discover Our promo codes Selected for you.