I was delighted to take part in a debate in the House of Lords highlighting the work of our Select Committee on AI. Our original report, AI in the UK: ready, willing and able?, was published in 2018 and made a large number of recommendations to the government. A follow up report was published in 2020 to consider what progress the government had made. In that period investment in AI had grown significantly and AI was being deployed in a wide range of fields: from agriculture and healthcare, to financial services, through to customer services, retail and logistics. However, the follow up report concluded that there was no room for complacency and made several further recommendations particularly around the ethical and appropriate use of public data.
The Committee Chair Lord Clement-Jones highlighted that AI lies at the intersection of computer science, moral philosophy, industrial education and regulatory policy. In my speech I picked up these themes through the practical policy demands of: data protection, the potential for bias in AI systems, the benefit of a distributed digital identity system and the pressing need for effective public engagement.
“It shows the foresight of the noble Lord, Lord Clement-Jones, in having a debate today because it is time to wish many happy returns— not to the noble Lord – but to the GDPR. Who would have thought that it is already four years since 25 May 2018?
In many ways, it has not been unalloyed joy and success. It is probably over-prescriptive, has not necessarily given more protection to citizens across the European community, and certainly has not been adopted in other jurisdictions around the world. I therefore ask my noble friend the Minister: what plans are there in the upcoming data reform Bill not to have such a prescriptive approach? What is the Government’s philosophy in terms of balancing all the competing needs and philosophical underpins to data when that Bill comes before your Lordships’ House?
Privacy is incredibly important. We see just this week that an NHS England AI project has been shelved because of privacy concerns. It takes us back to a similar situation at the Royal Free—another AI programme shelved. Could these programmes have been more effectively delivered if there had been more consideration and understanding of the use of data and the crucial point that it is our data, not big tech’s? It is our data, and we need to have the ability to understand that and operate with it as a central tenet. Could these projects have been more successful? How do we understand real anonymisation? Is it possible in reality, or should we very much look to the issue around the curse of dimensionalisation? What is the Government’s view as to how true anonymisation occurs when you have more than one credential? When you get to multiple dimensions, anonymisation of the data is extraordinarily difficult to achieve.
That leads us into the whole area of bias. Probably one of the crassest examples of AI deployment was the soap dispenser in the United States—why indeed we needed AI to be put into a soap dispenser we can discuss another time—which would dispense soap only to a white hand. How absolutely appalling, how atrocious, but how facile that that can occur with something called artificial intelligence. You can train it, but it can do only pretty much what datasets it has been trained on: white hands, white-hand soap dispensing. It is absolutely appalling. I therefore ask my noble friend the Minister: have the Government got a grip across all the areas and ways in which bias kicks in? There are so many elements of bias in what we could call “non-AI” society; are the Government where they need to be in considering bias in this AI environment?
Moving on to building on how we can all best operate with our data, I believe that we urgently need to move to have a system of digital ID in the UK. The best model to build this upon is the principles around self-sovereign distributed ID. Does my noble friend agree and can he update the Grand Committee on his department’s work on digital ID? So much of the opportunity, and indeed the protection to enable opportunity, in this space around AI comes down to whether we can have an effective interoperable system of digital ID.
Building on that, I believe that we need far greater public debate and public engagement around AI. It is not something that is “other” to people’s experience; it is already in every community and impacting people’s lives, whether they know it or want that to be the case.
We see how public engagement can work effectively and well with Baroness Warnock’s stunning commission decades ago into IVF. What could be more terrifying than human life made in a test tube? Why, both at the time and decades later, is it seen as a such a positive force in our society? It is because of the Warnock commission and that public engagement. We can compare that with GM foods. I make no flag-waving for or against GM foods, I just say that the public debate was not engaged on that. What are the Government’s plans to do more to engage the public at every level with this?
Allied to that, what are the Government’s plans around data and digital literacy, right from the earliest year at school, to ensure that we have citizens coming through who can operate safely, effectively and productively in this space? If we can get to that point, potentially we could enable every citizen to take advantage of AI rather than have AI take advantage of us. It does not need to be an extractive exercise or to feel alienating. It does not need to be put just to SEO and marketing and cardboard boxes turning up on our doorstep—we have forgotten what was even in the box, and the size of the box will not give us a clue because the smallest thing we order is always likely to come in the largest cardboard box. If we can take advantage of all the opportunities of AI, what social, economic or psychological potential lies at our fingertips.
What is AI? To come to that at the end rather than beginning of my speech seems odd. Is it statistics on steroids? Perhaps it is a bit more than that. AI, in essence, is just the latest tools—yes, incredibly powerful tools, but the latest tools in our human hands. It is down to us to connect, collaborate and co-create for the public good and common good, and for the economic, social and psychological good, for our communities, cities and our country. If we all get behind that—and it is in our hands, our heads and our hearts—perhaps, just perhaps, we can build a society fit for the title “civilised”.”
Lord Holmes of Richmond MBE, AI in the UK Debate, House of Lords, Grand Committee, 25 May 2022
We now look forward to the Data Reform Bill, announced in the Queens Speech, that will be coming before Parliament this session and which is a real opportunity to set standards and revive trust in a data protection regime that will support the use of AI systems for the public good. I hope also that the government will take seriously our calls for the ethical deployment of AI and the need for a public awareness campaign. With this technology, with these new tools, there is so much, almost limitless, potential but we must ensure the human is at the heart.