Asked by
To ask His Majesty’s Government what assessment they have made of
existing regulations and practices in relation to artificial
intelligence, and what plans they have to monitor and control
artificial intelligence (1) in the UK, and (2) in cooperation
with international partners.
The Parliamentary Under-Secretary of State, Department for
Science, Innovation and Technology () (Con)
The AI Regulation White Paper set out our proposed framework for
governing AI, including plans to establish a monitoring and
evaluation process to track performance. This will complement the
central AI risk function which we have established to identify
measures and mitigate risks. We work closely with international
partners through the G7, the GPAI and the Council of Europe to
understand AI risks, and are leading the way by convening the AI
Safety Summit in November.
(Con)
My Lords, I welcome the Government hosting the AI summit at
Bletchley Park, which is an opportunity to define the guard-rails
on the use and misuse of AI with international partners. AI is
borderless, as we know, so co-operation with others such as the
USA, China and the EU is vital. Given the advances in draft
legislation on AI by our neighbours in the EU, what plans do the
Government have to continue the co-operation and dialogue with
these other interests to give our thriving UK AI businesses
certainty in their ability to sell and trade into all
jurisdictions?
(Con)
My noble friend is absolutely right to highlight the essential
need for interoperability of AI given the way that AI is produced
across so many jurisdictions. In addition to the global safety
summit next week, we continue our very deep engagement with a
huge range of multilateral groups. These include the OECD, the
Council of Europe, the GPAI, the UN, various standards
development groups, the G20 and the G7, along with a range of
bilateral groups, including —just signed this year—the Atlantic
declaration with the US and the Hiroshima accord with Japan.
(LD)
My Lords, Professor Stuart Russell memorably said:
“There are more regulations on sandwich shops than there are on
AI companies”.
After a disappointing White Paper, in the light of the
forthcoming summit will the Government put more risk and
regulatory meat in their AI sandwich? Is it not high time that we
started addressing the AI risks so clearly identified at the G7
meetings this year with clear, effective and proportionate
regulation?
(Con)
I am pleased to say that the Government spend more on AI safety
than any other Government of any country. We have assembled the
greatest concentration of AI safety expertise anywhere and, based
on that input, we feel that nobody has sufficient understanding
of the risks or potential of AI at this point to regulate in a
way that is not premature. The result of premature regulation is
regulation that creates unnecessary friction for businesses, or
runs the risk of protecting or failing to protect from emerging
dangers of which we are as yet unaware.
(Lab)
My Lords, we learned again just this week that our own public
sector is already using this very powerful technology across the
board in Whitehall on matters such as criminal justice, health
and education, with great opportunity but great risk. Where is
the statutory framework for that current use of the technology?
At a time when so many of the Minister’s colleagues in the
Government want to walk away from international agreement, what
hope is there for us to deal with technology on a global scale
without new agreements, not fewer ones?
(Con)
I certainly do not recognise a situation in which many of my
governmental colleagues want to walk away from international
regulations; indeed, I have just provided quite a long list of
them. It is entirely appropriate that, within the bounds of
safety and their remit, different public sector bodies use this
crucial new technology. They do so not in an unregulated way but
with strict adherence to existing regulations.
(CB)
My Lords, can the Minister clarify how the Government intend to
regulate the use of NHS data, particularly the contract for its
collection, which is awarded to an overseas company? Furthermore,
the UKRI has requested that the Government invest in the
significant amount of computing power which we do not have but
require for generating AI in healthcare.
(Con)
The Independent Review of the Future of Compute, which we
accepted in its entirety, guided us to commit £900 million
initially to buying compute. We have confirmed the purchase of an
exascale system in Edinburgh as well as the UK’s soon-to-be most
powerful supercomputer, in Bristol. There will be further
announcements on this as part of the summit next week. The use of
NHS data is subject to not only stringent contractual
requirements but, already, stringent regulations about data
privacy.
(Con)
My Lords, does my noble friend agree that we need far greater
public engagement and public discourse around AI? Is he aware of
the alignment assemblies used in Taiwan to such good effect? Will
he consider taking a similar approach to such benefits in the
UK?
(Con)
I very much agree with my noble friend that we need maximum
public acceptance of AI. However, that must be based on its
trustworthiness. That is why we are pursuing, among other things,
the global AI Safety Summit next week. I am not familiar with the
Taiwanese approach but will look into it, and look forward to
discussing it in due course.
(Lab)
My Lords, it has been reported that the Government want big tech
companies to agree a set of voluntary guidelines at the AI
summit. Can the Minister confirm this? If so, why are the
Government not seeking more robust systems of oversight and
regulation, notwithstanding some of the advantages of AI, when
the dangers of unchecked technology are, as we have heard, so
high?
(Con)
I do not believe that anyone anywhere is advocating unregulated
AI. The voluntary agreement is, of course, a United States
agreement secured with the White House. We welcome it, although
it needs to be codified to make it non-voluntary, but that will
be discussed as part of the summit next week.
(CB)
My Lords, I would like to pick up on the point made by the noble
Lord, , because Professor
Russell also said that he would like to ban certain types of AI
deepfakes. With elections looming in this country, can the
Minister tell the House whether he thinks AI developers should be
banned from creating software that allows the impersonation of
people, particularly high-profile politicians?
(Con)
The noble Viscount raises an extremely worrying and serious
issue: the use of deepfakes to impersonate politicians. The
integrity of our entire political process could be placed at risk
with untrammelled and irresponsible use of these technologies.
However, I simply cannot see any pathway to banning these
technologies unilaterally, as where they are developed could be
absolutely anywhere on earth. I am afraid that any step we are
likely to take will not affect that.
Noble Lords
My Lords—
(Con)
It is the turn of the Liberal Democrat Benches.
(LD)
I thank your Lordships’ House. I will follow on from the point
made by the noble Lord, Lord Holmes. Huge commercial benefits are
possible from AI. We have talked about the dangers, but there are
benefits as well. However, as the Made Smarter Review made clear,
the management skills to implement the digital opportunities of
today are insufficient, so they are quite clearly not going to be
there to implement the benefits of the future. In conjunction
with his colleagues in the business department, what is the
Minister doing to make sure that we have the skills to be able to
take advantage of this technology?
(Con)
Yes, I thank the noble Lord for his point, which is a really
important one. There is no defined curriculum of skills for AI
anywhere, and there is such a very large range of different types
of skills from data science, analytics and computer science,
among others, to do that. I do not believe that anyone has
produced what might look like a core curriculum of those things.
We are, on the other hand, investing very serious funds into
education at all levels, from school age to college age and
advanced studies as well. I very much take the point, and driving
global acceptance and adoption of AI is absolutely key to
realising its value.
