UK needs its own ‘BritGPT’ or will face an uncertain future, MPs hear
The UK needs to support the creation of a British version of ChatGPT, MPs were told on Wednesday, or the country would further lose the ability to determine its own fate.
Speaking to the Commons science and technology committee, Adrian Joseph, BT’s chief data and artificial intelligence officer, said the government needed to have a national investment in “large language models”, the AI that underpins services such as ChatGPT, Bing Chat and Google’s Bard.
Without such technology, the nation would struggle to compete internationally in future, he said.
“We think there’s a risk that we in the UK, lose out to the the large tech companies, and possibly China, and get left behind … in areas of cybersecurity, of healthcare, and so on. It is a massive arms race that has been around for some time, but the heat has certainly been turned up most recently.”
Dame Wendy Hall, who co-chaired the UK government’s AI review in 2017, concurred with the need to develop a BritGPT. “If we don’t do it, we just become a service industry country,” she told MPs. “But in the UK, we can harness the technology, use that to drive the economy and grow jobs.”
The computing power required to perform cutting-edge AI work is expensive, MPs were told, which prevents the UK’s leading researchers in the field from competing directly with large, well-funded US companies.
“University researchers are at risk of being left behind,” said Nigel Shadbolt,
the chair of the Open Data Institute, “because their access to the kinds of [computing power] you need is not organised terribly systematically. We’ve got to think about we can sustainably guarantee our access to that.”
Training GPT-3, the language model on which ChatGPT is based, took about $10m-worth of computing power at public prices in 2020, according to OpenAI’s paper announcing the technology. Improving it to the level of ChatGPT, released in December 2022, will have taken millions of dollars more, with even more expenditure for the human “raters” who trained it to respond well to the Q&A format.
A report published on Wednesday from the Tony Blair Institute, co-authored by the former Labour prime minister and his one-time Conservative rival William Hague, also called for the same investment.
“Given these AI systems will soon be foundational to all aspects of our society and economy, it would be a risk to our national security and economic competitiveness to become entirely dependent on external providers,” the paper argues.
“Since the technology is sufficiently mature, the government should take on a greater role in its direct development to ensure the UK has sovereign capabilities in this field. Leading actors in the private sector are spending billions of dollars developing such systems so there may only be a few months for policy that will enable domestic firms and our public sector to catch up.”
The Blair institute report argues that such a “sovereign general-purpose AI capability” should be supported by a direct investment into supercomputing infrastructure, some of which should be specifically reserved for training those large AI models, with a long term goal of treating it as a utility “much like our water or energy systems”.