Researchers in the UK have urged that generative AI systems must be leveraged to provide a major boost to the British economy.
According to a new University of Cambridge report, the UK lacks the computing capacity and capital needed to build generative AI technology fast enough to compete with leading US companies like Google, Microsoft, and OpenAI.
Instead, the UK must leverage generative AI through real-world applications, such as developing new diagnostic products and addressing the shortage of software engineers.
The UK Government should help in this endeavour via offering tax breaks for businesses developing AI-powered products and services.
However, new legislation must also be brought in to ensure that the UK has solid legal and ethical AI regulation. Without this, British industries may struggle to trust emerging AI platforms, such as ChatGPT, enough to invest time and money into it.
The policy report is a collaboration between the university’s flagship on Artificial Intelligence, ai@cam, Cambridge’s Minderoo Centre for Technology and Democracy, and Bennett Institute for Public Policy.
What is generative AI?
Generative AI uses algorithms trained on giant datasets to output original text, audio, video, or images at a high speed and large scale.
Examples of the technology include ChatGPT, which has dominated the headlines this year, and Midjourney, which can create an image in any style in seconds.
“Generative AI will change the nature of how things are produced, just as what occurred with factory assembly lines in the 1910s or globalised supply chains at the turn of the millennium,” said Dame Diane Coyle, Bennett Professor of Public Policy.
The technology requires Graphics Processing Units
Graphic Processing Units (GPU) are networked grids of computing hardware. These clusters are used to handle the vast quantities of data that hone generative AI.
ChatGPT alone is estimated to cost $40m a month in computing costs.
Earlier in the year, the UK chancellor announced £100m for a Frontier AI Taskforce to scope out the creation of UK AI technology.
However, the researchers argue that the supercomputer announced by the chancellor is unlikely to be online until 2026. As well as this, none of the big three US tech companies, Google, Amazon, and Microsoft, have GPU clusters in the UK.
“The UK has no companies big enough to invest meaningfully in foundation model development,” said report co-author Sam Gilbert.
“State spending on technology is modest compared to China and the US, as we have seen in the UK chip industry.”
The UK should leverage its strengths to harness the everyday uses of AI
As the UK does not have the industry to invest in the foundational development of generative AI, the report states that the region should use its strengths in fin-tech, cyber security, and health-tech to build software that leverages AI for everyday use.
“Generative AI has been shown to speed up coding by some 55%, which could help with the UK’s chronic developer shortage,” said Gilbert.
“In fact, this type of AI can even help non-programmers to build sophisticated software.”
As well as this, the UK has research universities that could drive progress in tackling AI challenges. For example, the cooling of data centres to the detection of AI-generated misinformation.
New responsible AI legislation is required
However, UK organisations lack incentives to comply with responsible AI.
“The UK’s current approach to regulating generative AI is based on a set of vague and voluntary principles that nod at security and transparency,” said report co-author Dr Ann Kristin Glenster.
“The UK will only be able to realise the economic benefits of AI if the technology can be trusted, and that can only be ensured through meaningful legislation and regulation.”
The report argues that as well as new legislation, a series of tax incentives should be introduced to increase the supply of capital to AI start-ups and businesses including generative AI in their operations.
Bottom-up uses of generative AI within organisations could be identified with challenge prizes.
