TECH NEWS – A Japanese-American theoretical physicist has made some pretty intense criticisms of AI chatbots and large language models (LLMs).

Fear of the unknown is natural, and the fear of artificial intelligence becoming evil has been implanted in us through the Terminator and Matrix films. But perhaps we shouldn’t prepare for the worst-case scenario: Michio Kaku, professor of theoretical physics at the City College of New York and CUNY Graduate Center, considers current AI models more than glorified tape recorders!

CNN’s Fareed Zakaria interviewed the professor. Kaku’s logic makes sense: AI takes the elements that someone has put on the Internet, stitches them together, and then delivers the results as if AI created them, and the audience goes wild thinking how humanlike it is, but the professor says the technology can’t tell the truth from the falsehoods. So if you tried ChatGPT or Google Bard and they gave you false information in response, that’s why it happened.

An LLM is an extensive database that gathers data from the Internet, and then statistical models and algorithms analyze it all to generate humanlike, probability-based answers to our queries. But a lot of data requires a lot of computing power, and it’s no accident that Nvidia, for example, is making a killing with its products. LLMs are a good fit for quantum computers because they can process parallel data. If we get to working quantum machines, the artificial intelligence industry could take an indescribably giant leap forward.

The interview embedded below talks mainly about quantum technology (Professor Kaku also explains how it works), and the interviewee perhaps not coincidentally thinks that the AI industry will be overtaken by quantum. Still, the title is no coincidence either (“The next evolution in tech needs to be nearly frozen to operate”)…

Source: PCGamer