Nick Clegg insists that many AI risk warnings do not apply to the current generation of bots. Photo: KENZO TRIBOUILLARD/AFP via Getty Images
Artificial intelligence (AI) technology. too «stupid» to threaten humanity, Sir Nick Clegg said when Meta introduced its rival ChatGPT.
The former vice premier, who is now Meta's head of global affairs and communications, downplayed warnings about the threat of AI, saying the «existential warnings» refer to technologies that «do not currently exist.»
His comments came after Meta, which owns Facebook and Instagram, said it would make blueprints for its new AI available to businesses for free.
The AI company's chatbot, called Llam a, will be «open source», which means that organizations will be free to inspect or copy its data. for his own ends.
Sir Nick said on BBC Radio 4's Today: «I think the hype is somewhat ahead of the technology.
“I think a lot of the existential warnings are about models that don’t currently exist, the so-called super-intelligent, super-powerful AI models — a vision in which AI develops autonomy and agency, where it can think for itself and reproduce itself.”
“The models that we open source are far from that. In fact, they are pretty dumb in many ways.”
0605 AI Impact
Meta is the latest tech giant to launch an AI-powered chatbot. These text-generating chatbots use machine learning to convincingly answer questions, compose emails, or write essays with human-like answers.
Meta is set against OpenAI, the Silicon Valley startup behind ChatGPT, which took the tech world by storm last year with its powerful AI-powered chatbot.
Meta's decision to make its AI code widely available potentially undermines competitors who charge companies to use their technology.
However, some experts fear that Meta's decision to open up its technology could expose it to attackers or cybercriminals.
The company's terms insist that its AI cannot be used to promote violence, create computer viruses, create weapons or nuclear technology. It also prohibits spamming, hate speech or child abuse.
Sir Nick insisted that his model was «safer» than the competition and was fined hundreds of attempts to hack it. Asked about the risks the models pose, Sir Nick said they «can't build a nuclear bomb.»
He added: «It's not like we're at a T-junction where firms can choose whether to open source or not. Models are already constantly in the public domain.”
Mark Zuckerberg's meta has decided to open source its artificial intelligence bot Llama. Credit: Kevin Dietsch/Getty Images
Meta technology will be language-centric, so it cannot be used to create illegal images or «deeply fake» pornography.
Sir Nick added that most current AI tools are not as effective as claimed.
He said his tools work like automatic text generators that «guess at great speed,» but they have «no innate autonomous intelligence at all.»
Meta Lama will be available online to businesses as part of a deal with Microsoft, the tech giant that has also invested billions in a system that competes with OpenAI.
Mark Zuckerberg, founder and CEO of Facebook, said that the discovery of his models «has increased safety and security, because when software the software is open, more people can scrutinize it.”
Some experts warn that the pace of AI development threatens humanity. Thousands of industry figures and researchers have signed a letter demanding a six-month halt to new developments.
Billionaire Elon Musk, who runs his own AI research company, warned of the risk of a «terminator future,» a reference to a series of films in which humanity is virtually wiped out by machines.
Свежие комментарии