MPs returned to the House of Commons yesterday. Amid a packed programme ranging from the Ukraine to the Middle East and the recent riots in the UK, were dozens of maiden speeches. Some of these came during a debate on Technology in Public Services, led by Secretary of State Peter Kyle.
Unsurprisingly, this was one of the first debates to follow the Summer Recess. Since the General Election, we have heard a lot about the potential for AI to transform public services and deliver efficiencies for the public purse. We’ve also seen machinery of government changes aimed at turning the Department for Science Innovation and Technology (DSIT) into the digital centre of government, with the Government Digital Service (GDS), Central Digital and Data Office (CDDO) and the Incubator for AI all now under the DSIT roof. Also, in the past few weeks, an AI Action Plan has been ‘in the works’, led by Matt Clifford, the VC and former architect of the AI Safety Summit.
The ODI has a long-held position that benefits can’t be realised without good, well-governed data, responsible data stewardship, or mitigating potential harms. Our recent work on data-centric AI reveals that issues, including a shortage of high-quality data and inadequate data governance, threaten to derail any benefits that might be gained from the current technology revolution. As our Executive Chair and co-founder, Sir Nigel Shadbolt, has said:
...if the UK is to benefit from the extraordinary opportunities presented by AI, the government must look beyond the hype and attend to the fundamentals of a robust data ecosystem built on sound governance and ethical foundations. We must build a trustworthy data infrastructure for AI because the feedstock of high-quality AI is high-quality data. The UK has the opportunity to build better data governance systems for AI that ensure we are best placed to take advantage of technological innovations and create economic and social value whilst guarding against potential risks.
This message has formed the backbone of our recommendations to the team developing the AI Action Plan. Here, we have focused squarely on the data layer, and government plans for a National Data Library. Developing a robust data infrastructure is fundamental to the Action Plan's success.
We’ve also centred on areas where we feel we can bring the most value and have the greatest expertise - in data governance, ethics and data-centric AI. You can read a summary of our recommendations.
Indeed, many of the issues highlighted in our recommendations were strongly featured in last night’s Commons debate. It was also interesting to hear how some of the concerns raised by ODI’s network at our policy roundtable in June are now being played out in Parliament.
Trust, transparency, and accountability
Dawn Butler (Lab) asked the Secretary of State to reassure the public that organisations would fully understand their responsibilities in terms of public safety. He agreed, saying: “We need to reassure people that as we embrace the technological advances that sit before us, we do so in a way that has safety built in from the outset.”
At our roundtable, participants made this one of their central areas of concern, emphasising the importance of trust, transparency, and accountability in data use. And in our recommendations for the AI Action Plan, we focus firmly on public trust and transparency, advocating for clear communication and governance to ensure confidence in AI technologies.
Cybersecurity
Our roundtable participants also raised the importance of strengthening cybersecurity measures to protect against malicious attacks. Last night, Max Wilkinson (Lib Dem) emphasised the importance of cybersecurity as digital infrastructure becomes integral to public services. In our recommendations for the AI Action Plan, we call for robust governance structures within the National Data Library to secure data and maintain public trust.
Digital inclusion and accessibility
Last night’s debate focused on multiple areas of digital inclusion, including ‘the digital divide’ and ensuring equitable access to technology. Jim Shannon (DUP) highlighted concerns for elderly citizens who are often unable to access digital services, raising the possibility that inequities in public services could become further entrenched.
Chi Onwurah (Lab) made a case for greater diversity in hiring for tech roles in the civil service, noting that "the private sector has driven the adoption of so many of the technologies that have transformed our lives, but we must build state capacity in tech by recruiting more diverse science and technology experts to the civil service so that government can innovate."
The ODI is a strong proponent of people-centred technology and ensuring that all sectors of society are represented in design, decision-making and delivery. We advocate for a user-centric design in the National Data Library to ensure that digital services are accessible, and we have long argued for participatory approaches to data and technology development. We hope this is strongly represented in forthcoming policies, including the AI Action Plan.
Ethics and bias in AI
Data ethics has long been an area of focus for the ODI. During our June roundtable, our network strongly stressed the need for statutory AI principles and ethical AI deployment. Indeed, this is particularly important for AI in public services. Ensuring that the data on which AI is trained is - as far as possible - diverse and unbiased is especially important so that everyone benefits equally and is not disadvantaged by technological advancements. In last night’s debate, Debbie Abrahams (Lab) and Layla Moran (Lib Dem) highlighted the potential risks of bias in AI. They called for ethics to be carefully considered, the latter stating that “AI is not without its challenges. It reflects the values and biases of the humans who create it, causing campaigners to raise concerns.”
Unsurprisingly, this is a key theme in our AI Action Plan response, where we emphasise the need for responsible AI practices to ensure ethical data use and prevent the exacerbation of inequalities.
Investment in AI and digital infrastructure
The Secretary of State was robust in replying to various Conservatives, including shadow DSIT secretary Andrew Griffith, raising the £1.3bn cut to AI and digital infrastructure, including shelving plans for a £800m exascale supercomputer at Edinburgh University. In response to a question from David Reed (Con) he said he could not cut “something that did not exist in the first place.” Instead, he pointed to plans to bring forward “binding regulations on the handful of companies developing the most powerful AI systems.”
Peter Kyle specifically mentioned the importance of data underpinning the roll-out of safe AI, saying that digital infrastructure and data will be “the driving force behind Britain’s digital transformation, better hospitals and schools, safer streets and transport that works for working people.”
However, hearing more mention of data as infrastructure would have been good. The ODI has long argued that data needs a strong infrastructure to underpin its collection, use and sharing. This goes beyond the technical into ethics, governance and stewardship.
We look forward to seeing increased investment in the UK’s data-centric AI ecosystem to ensure effective and responsible AI use.
Efficiency and reducing bureaucracy
Peter Kyle also discussed how AI can reduce the bureaucratic burden of public services. Indeed, in our AI Action Plan recommendations, we agree that AI has the potential to enhance public services significantly. However, “achieving such efficiency requires high-quality data and robust management and governance practices.” To deliver this, we propose a modular, decentralised data infrastructure to streamline and improve services while reducing inefficiencies. Attempts to reform public services should emphasise the delivery of innovative, high-quality services that are well-targeted to those who require them; the efficiency argument should not simply be about saving money but also about making things better. Strong data infrastructure can enable both - but it is not cost-neutral.
Cross-government collaboration and governance
Gareth Snell (Lab/Co-op) emphasised the importance of cross-government collaboration in implementing digital technologies. For the National Data Library to deliver social and economic benefits, it must be an exemplar of integrated and accessible data infrastructure across government.
It remains to be seen whether the new organisational structure of government departments handling data and AI and their roles and responsibilities will deliver a more collaborative and - overall - better approach. As Andrew Griffiths (Con) pointed out, only “time will tell”.
Public awareness of data and AI
In our June roundtable, people felt that increasing public awareness and understanding of data and AI is a priority. This is another area in which we’d like to see a more comprehensive consideration of participatory approaches to decision-making and delivery of public services. This is also a priority for the ODI, aligning with our focus on public trust and transparency, emphasising the need for clear communication and public engagement to build trust in data and AI technologies.
Indeed, it was refreshing to see in the debate a focus on people throughout, individuals as citizens and consumers. Chi Onwurah (Lab), Allison Gardner (Lab), Natasha Irons (Lab) and Layla Moran (Lib Dem) all had something to say on this topic. The Secretary of State put it thus: “It is all too easy to think of technology such as AI as being impersonal, alienating or distant, but the first thing I think about is people—the teachers in our schools who will deliver a personalised lesson to every pupil and help them fulfil their potential, and the patients in our hospitals who can access lifesaving drugs for diseases that until recently were untreatable. Technologies can change our everyday lives in ways that are both ordinary and extraordinary.”
We look forward to hearing how the government plans to bring more people’s voices into the delivery of technology policy and to seeing our recommendations considered as part of the AI Action Plan.
