The rapid growth of advanced artificial intelligence, spurred in part by OpenAIs chatbot ChatGPT, pushed some tech industry leaders last week to call for a six-month ban on further development of the technology. Apple cofounder Steve Wozniak and Tesla chief Elon Musk were among those who signed the letter supporting the temporary suspension.
But ex-Google CEO Eric Schmidt, who has previously warned about the risk of using A.I., said that he wouldnt support a pause in its development. His reason? A pause would help other countries gain an edge.
Im not in favour of a six-month pause because it will simply benefit China, Schmidt told the Australian Financial Review on Thursday. What I am in favour of is getting everybody together ASAP to discuss what are the appropriate guardrails.
The proposed moratorium asks companies to stop development of complex A.I. systems for six months to address concerns over misuse and the profound risks to society and humanity. Some leaders, such as Alphabet CEO Sundar Pichai, have said they wont comply with the call to pause, stating that he supports regulation rather than a hiatus.
According to Schmidt, we may not be able to make sense of the threats from A.I. just yet which could pose a challenge in itself.
I think the concerns could be understated. I think things could be worse than people are saying, he said. You have a scenario here where you have these large language models that as they get bigger, have emergent behaviour we dont understand, referring to companies in the tech sector that are working on A.I. advancements.
Schmidts remarks come at a time when tools like ChatGPT and Midjourney, an A.I. image-generator, have set off an A.I. revolution worldwide. Companies like Google, Meta, and Microsoft have been scrambling to get a read of whats next in the A.I. frontier, as have Chinese companies like Baidu. Ernie Bot, a chatbot akin to ChatGPT, was a let-down for Chinese investors when it was launched in March, leading to Baidu losing $3 billion in market value. But that hasnt stopped the company subsequently improving its chatbot capabilities.
It is not perfect yet, so why do we need to launch it today? Baidu CEO Robin Li said at the launch event. Because the market demands it.
A.I. alarm bells
Part of the appeal (and concern) about A.I. is that its been widely adopted across the boardby the average person for writing essays and planning vacations, as well as by large financial institutions to support professional roles such as helping financial advisors do research. And as A.I. platforms become more commonplace, Schmidt, like other experts, has voiced concerns about the need for regulations in this space.
I think today the governments response would be clumsy because there are very few people in government who understand this stuff. So Im in favour of letting the industry try to get its act together, Schmidt said, suggesting that businesses create their own voluntary guidelines. This is a case where you dont rush in unless you understand what youre doing.
One of Schmidts biggest concerns is that A.I. systems become more complicated than people can fully comprehend, but companies, nevertheless, continue to introduce new products.
You can get [into] a situation where its a race to the bottom and we rush to release without knowing what were doing, he told the AFR.
I think the concerns could be understated. I think things could be worse than people are saying, Schmidt said. You have a scenario here where you have these large language models that as they get bigger, have emergent behaviour we dont understand.
The former Google chief is also worried about the potential exploitation of A.I. in things like cybercrime if safeguards arent imposed. In recent weeks, fake A.I.-generated images of the Pope in a puffer jacket have been making the rounds online, showing people how good and dangerous A.I. has become at the same time. There have also been growing concerns about A.I. text tools being used to plagiarize work in schools and universities.
Schmidt, who co-authored a book about the A.I. revolution in 2021, has repeatedly cautioned that if the industry lacks frameworks for how to work with this technology, it could have dire consequences on society and the democratic system. But he has also acknowledged the benefits of using A.I. to upgrade antiquated technologies in the military and helping improve education.
Imagine a world where you have an A.I. tutor that increases the educational capability of everyone in every language globally. These are remarkable, Schmidt said in an interview with ABC News last week.
