AI will be Extremely Unpredictable and Unimaginable, warns I…
Analytics India Magazine (Siddharth Jindal)
AI is going to be both extremely unpredictable and unimaginable, said Ilya Sutskever, founder of Safe Superintelligence, in a recent video interview.
He said that the development of advanced AI systems could lead to a point where AI begins to improve itself. He warned that this could trigger rapid and uncontrollable progress, making it difficult for humans to understand or manage what comes next.
He was optimistic about the concept that an AI could create the next generation of AI, referring to the intelligence explosion and questioned how one should respond to that possibility.
“If the AI became capable enough, we’ll have incredible health care. You could cure so many diseases and maybe extend life,” he explained on the potential of AI. Yet, he paired this optimism with a stark warning.
Sutskever traced his journey from self-taught student to AI pioneer while accepting an honorary degree from the Open University. His first encounter with advanced learning began in the eighth grade.
“I just got those books and I read them and I understood,” he said. “I became confident that if I just read something very slowly, I would eventually understand it.”
After moving to Toronto, Sutskever chose not to finish high school and instead transferred directly into the University of Toronto. He called it “the place to be” after discovering Geoffrey Hinton was on the faculty.
“How is learning possible at all?” he recalled wondering. “Can computers learn? Somehow, I felt that if you had the answer to the learning question, then everything else would follow.”
That question drove his work on the landmark AlexNet paper during his graduate studies. He said the response to the research was swift, with companies interested in acquiring “our company, which didn’t exist,” prompting him and his colleagues to create one. The venture was later acquired by Google, where Sutskever continued his research on large neural networks.
He said the decision to co-found OpenAI came from a desire “to try a real serious startup with all these illustrious people.”
In closing, Sutskever said the honorary degree “closes a circle,” as the Open University had once represented “all of interesting learning” in his life.
The post AI will be Extremely Unpredictable and Unimaginable, warns Ilya Sutskever appeared first on Analytics India Magazine.
Generated by RSStT. The copyright belongs to the original author.