Tony Blair is wrong – AI will not magically solve our public services

Too many people might be left out of its revolution.

Tony Blair was well known for being more of a notemaker than a typist, and confessed early on during his tenure as prime minister that when he did use a computer, he would “usually need help”. How times have changed. These days Blair evangelises technology and AI with the zeal that can only come from conversion. He is regularly on the airwaves calling for AI to be unleashed on public services to transform education, health and welfare – and he is seen as influential on the new government’s thinking. Undoubtedly this technology is developing fast and could play an important role in people’s lives, but those who are too starry-eyed about AI often overlook a range of issues.

Public services are complex systems, and rushing to bolt on unproved technology is unlikely to work. The UK does not have a good track record here and the danger is we will see the same kinds of IT transformation project failures that have been commonplace over the years in the public sector. In any nascent technology, and especially one as expensive as AI, the government will need to ask for much higher-quality evidence of costs and benefits. There is a lot of snake oil for sale.

It hardly needs saying that public services are in a poor state, and this is especially true in computer systems. A few years ago, thousands of NHS computers were found to be running Windows XP, which was supposed to be phased out in 2014. Datasets across government could be linked more effectively; the underlying infrastructure will need overhauling if any benefits are to be realised. If it takes the hype of AI to give political cover to this kind of important investment in basic IT and data systems, then so be it.

At the same time, however, AI needs to work for the people who are worst off in society. Across the UK, there are huge disparities in access, digital skills, usage and outcomes. Among people living on household incomes under £25,000, one in five never use the internet. This rises to nearly a third of disabled people and nearly half of those aged 65 or over. The very people whose lives a Labour government would wish to transform through public services are the ones most likely to be passed over by the AI revolution. It is likely they will not be well represented in the data that algorithms are trained on, so the needs of the poor or disabled won’t be well embedded in AI models.

There is also a risk that the vulnerable are at the sharp end of the AI revolution when it comes to managing their interactions with the state. During a period of tight public finances, digitisation can sometimes be seen as a code word for cost-cutting: let’s lose the messy interactions with citizens and get them to deal with us through an app. But it must not be blindly trusted; when the Department for Work and Pensions used an algorithm to flag high-risk benefit claims, two-thirds of those flagged turned out to be legitimate. If things go wrong, as we’ve seen with the Post Office scandal, it can be hard for people to get fairness and accountability from the system in such circumstances. High-stakes decision-making seems a bad place to start with AI – it might be wiser to start by automating dull, back-office processes, of which there are plenty to fix in the health service and local government. And rather than rushing breathlessly to unleash technology, let’s instead have close consultation with users of public services to improve their design and see how AI can give them what they want and need.

Also messianic about the transformative potential of AI are – surprise! – the large technology companies who sell it. The market is concentrated and there are considerable dangers of locking in to a provider that you cannot leave, as they run the platform for your services. Thankfully Peter Kyle, the Science and Technology Secretary, has recognised the power of the tech giants, rightly saying that negotiating with them is as political as managing diplomacy with states.

It’s a moment for the rise of a new and unlikely star in government: the procurement manager. While AI may seem like magic, in fact much of its power is in the data that algorithms are trained on. Procurement managers need to recognise the very significant value of data the state has amassed, and which the tech titans are hungry for. This provides negotiating power – it should not just be given away. Those in charge of making purchasing decisions might also band together so not every school, local authority or NHS trust is having to face off against the might of large tech companies alone.

The British Academy has been exploring ideas for what makes for a “good digital society”. This helps us ask how we might use technology to create a more relational set of public services, with a stronger voice for users and a focus on strengthening social and democratic outcomes. At this pivotal moment, civil society needs to be brought to the table to provide a wider range of perspectives and help safeguard the public good. Technology has the power to create a better society, but it’s unlikely to happen if AI is unleashed upon us. Taking the time to do things right isn’t anti-innovation – it is what is required to get deep-rooted change in social systems.