There are two opposing views about artificial intelligence: It could either save the world or plunge it into dystopia. And the CEOs of companies leading the charge on A.I., it turns out, are unsure themselves.
After years of slow and steady research, the A.I. wars are heating up as tech giants and startups alike rush their products to market. OpenAI opened the doors in November with the launch of its wildly popular ChatGPT, and established tech firms including Google, Microsoft, and Chinese search engine giant Baidu plan to launch their own advanced chatbots soon.
While intelligent chatbots like ChatGPT still make mistakes and are not quite as world-changing as believers like OpenAI founder Sam Altman eventually aspire them to be, it might be the first step towards A.I.’s commercialization and its mainstream integration into our lives. As Jeremy Kahn put it in Fortune’s most recent magazine cover story, ChatGPT’s debut last year could be artificial intelligence’s “Netscape Navigator moment,” a product release that instills a technology in the zeitgeist and predicates the arrival of a massive industry.
If A.I. is going to play a role in more aspects of our lives, it will be critical to ensure it doesn’t reach a runaway state, a hypothetical scenario in which A.I. gains above-human level intelligence and dominates its creators. It’s a risk CEOs like Microsoft’s Satya Nadella are well-aware of.
“Runaway A.I., if it happens, is a real problem,” Nadella said in an interview with CBS Mornings published Wednesday, echoing similar words of caution from OpenAI’s Altman and Elon Musk. But Nadella is still optimistic about A.I.’s chances to deliver on its promise, and humans may not even lose their jobs—as some fear will happen—to the increasingly intelligent technology, since a big part of ensuring that A.I. remains under control is for people to carefully guide it.
“The way to sort of deal with [runaway A.I.] is to make sure it never runs away,” Nadella said.
‘Lights-out’ or ‘utopia’?
As far as Nadella is concerned, the path A.I. and humanity are currently on points towards a better future for society. “It’s utopia,” he told CBS when asked whether A.I. is taking us to a better future or a bleaker one.
It’s a more certain and optimistic view than OpenAI’s Altman, who despite his company’s progress over the past few months—including a $10 billion investment from Microsoft—is measured when asked about his vision for A.I.’s future.
“I think the good case [for A.I.] is just so unbelievably good that you sound like a crazy person talking about it,” Altman said during an event for venture capitalists in San Francisco last month, but added: “I think the worst case is lights-out for all of us.”
If A.I. were to become a danger to us, it would be because humans let it happen, Nadella suggested during his interview with CBS. A.I. products should only be implemented in areas humans already have a solid grasp on and that humans can take over if things start going wrong, he said. Basically, we shouldn’t trust A.I. to do things we wouldn’t do ourselves.
“Before we even talk about alignment and safety and all these things that one should do with A.I., let’s talk about the context in which A.I. is used,” Nadella said. “The first set of categories in which we should use these powerful models are where humans unambiguously, unquestionably, are in charge.”
While A.I. can churn out detailed texts and information, Nadella insisted not only are our livelihoods safe, but some jobs could become less tedious because of new models like ChatGPT and those being developed at Microsoft.
“What this does really is create the draft. But the draft has to be read by somebody, has to be edited by somebody, approved by somebody,” he said. “I believe it creates more, I’ll call it both satisfaction in current jobs and net new jobs.”
Microsoft this week unveiled new versions of its Bing search engine and Edge browser, both of which have long failed to gain traction against rival Google. But the newest iteration of Microsoft’s products will incorporate an A.I. assistant language model similar to ChatGPT that will be incorporated into Bing.
The new model is not yet available publicly; there is a waitlist for it. But Nadella is impatient for a wide release, despite the risks that come along with releasing a largely untested product.
“The only way for any new technology to be really perfected is to be in the market with real human feedback, particularly with A.I.,” he said.
