Geoffrey Hinton's Lecture on AI Risks and the Future

Geoffrey Hinton, known as the godfather of artificial intelligence, delivered a lecture with Russian subtitles and a Q&A session from the audience. In his speech, he touched upon key aspects of the development of modern neural networks and their impact on society.

Hinton explained the principles behind ChatGPT and discussed the advantages of digital intelligence over human intelligence. He expressed the opinion that AI models should not have been released to the public without adequate safety checks.

Particular attention was given to the risks associated with technological advancement. The speaker described how AI might develop an instinct for self-preservation at any cost, and why the US and China would be unable to agree on the creation of combat robots.

Hinton also shared his personal views on the future interaction between humans and machines: in his opinion, AI should become not just an assistant, but a reliable support system, akin to a mother. He refuted accusations of panic-mongering, emphasizing the importance of conscious actions by developers.

The discussion covered complex philosophical questions: the presence of consciousness in AI, the possibility of genuine emotions in neural networks, and the problem of inner experience. The speaker noted that even humans lack a true "inner world," which complicates the task of creating genuine intelligence.

Geoffrey Hinton is a recipient of the Turing Award, the Nobel Prize in Physics (2024), and the Queen Elizabeth Prize for Engineering. He worked at Google for 10 years, becoming a Vice President, but left in 2023 to openly warn about the dangers of AI.

His great-great-grandfather, mathematician George Boole, created Boolean logic, which underpins the operation of all modern computers. Hinton's student, Ilya Sutskever, became a co-founder of OpenAI. The full text of the lecture is available in English.