Google Deepmind Co-Founder talks about Future AI Technologies
In his opinion, future AI developments will become more dynamic and will be able to act independently.
"You just give them a high-level common goal, and they will use all available tools to achieve it. Neural networks will talk to other people and AI," explained the head of Deepmind.
With sufficient freedom of action, new technologies will provide more functionality than existing generative algorithms, which Suleiman called "static".
Earlier, in an interview with CNBC, the programmer predicted that over the next five years everyone will be able to have personal assistants with artificial intelligence. His company, Inflation AI, launched its Pi chatbot in May as a competitor to OpenAI's ChatGPT. The tool focuses on personal advice and communication.
"The first wave of AI concerned classification. Deep learning has shown that we can teach a computer to distinguish between different types of input data: images, video, audio, language. Now we are in a generative wave, when you take information and create a new one. The third wave will be interactive," Suleiman added.
Speaking about the risks from AI, the co-founder of Deepmind said that "people will always be in command." For greater control, countries should create appropriate legislation, the developer believes.
"In general, there are certain points that we should treat very carefully, if not just exclude them in the foreseeable future. Something like recursive self-improvement. You wouldn't want your little AI to shut down and update its own code without your control," he said.
Suleiman summed up that the world has done an amazing job regulating "super-complicated things" like aviation, so it will be able to cope with artificial intelligence.
Recall that in August, YouTube CEO Neil Mohan shared plans for the introduction of AI in the music industry. He noted the rapid development of technology and spoke about the principles of the platform for its integration.