Academic research is needed to give machine learning an “ethical foundation,” according to the head of the AI engineering team at Intel, one of the world’s largest semiconductor chip makers.
Speaking at the Digital Universities UK event held by Times Higher Education Wei Li, vice president and general manager of Intel’s artificial intelligence and analytics division in partnership with the University of Leeds, said: You get what you get, and you don’t really know why.
He added: [such as healthcare], you’d want to know why that system gave that answer. ”
Academic research could help clarify fundamental ethical issues in AI, such as built-in biases about gender, Dr. Li said at the event.
“There are so many unknowns and unsolved problems in machine learning today that we really need basic research that universities can do but industry cannot,” he said.
Dr. Lee said his team at Intel is working on technologies that will help create fair and inclusive AI systems.
but speak with The, he warned that the industry is focusing more on building AI systems than on addressing the ethical issues they pose. he said. “Those things – I hope people in academia can do something deeper than what we do.”
A recent open letter signed by AI experts and industry executives, including Elon Musk, called for a moratorium on AI development until “we are confident that the impact of AI will be positive and the risks manageable.” But Dr. Li doesn’t expect it. AI progress slows down.
“It’s not realistic,” he said. “It’s a commercialization risk, a race to be the fastest and first in the industry. That’s enough incentive for people to want these things.”
So with rapid advances in AI systems, will academic research on machine learning models quickly become obsolete? No, as Dr. Li argued, universities could influence how future systems are built. said. “I don’t expect them [researchers] Digging into ChatGPT and explaining ChatGPT is not possible with the current state of the art. But with a better foundation of machine learning, the next generation could be a safer and less biased model. ”
Regarding university education, Dr. Li said: Institutions should teach students to be “more than just Messengers of ChatGPT,” he added.
“The outcome of college is the students you are producing. If they are not better than ChatGPT, why should you send them to college?” Dr. Lee asked.
sara.custer@timeshighereducation.com
