AI language systems are ‘quite stupid’ said Nick Clegg
Facebook-owner Meta’s president of global affairs Nick Clegg described current Artificial Intelligence (AI) models as “quite stupid”.
“The hype has somewhat outpaced the technology”, said the former UK deputy prime minister.
According to him, current models fall far short of warnings when AI develops autonomy and thinks for itself.
He told the BBC’s Today Programme that they are quite stupid in many ways.
Meta announced that its large language model, Llama 2, would be open source and free for everyone to use.
A large language model – the technology that powers chatbots like ChatGPT – works by joining dots in huge datasets of text, and guessing the next word in a sequence. AI experts are warning about existential threats related to systems that don’t yet exist, he said.
Meta’s decision to make Llama 2 available for commercial businesses and researchers has divided the tech community. Llama, the first generation, has already been leaked online within a week of its release.
Open-source is a well-trodden path in this sector – allowing others to use your product gives you free user testing data, helping you identify bugs and other problems.
However, whatever Sir Nick might say, this is an extremely powerful tool.
It is well known that previous versions of chatbots have been manipulated to spread hate speech, generate false information, and give harmful instructions. Is Llama 2’s guardrail strong enough to prevent misuse in the wild, and what will Meta do if that happens?
Also of note is Meta’s decision to partner with Microsoft on this – Llama 2 will be accessible and usable via Microsoft platforms such as Azure – as Microsoft has also invested billions of dollars in ChatGPT creator OpenAI.
The company has its sights firmly set on artificial intelligence, and it has the deep pockets to buy its way into the market. There is a risk that the AI pool could soon become dominated by a few very large fish – and is that healthy for competition?