Centre for Data Ethics and Innovation

Lord Clement-Jones Excerpts

Wednesday 21st November 2018 Lords Chamber

[Lord Stevenson of Balmacara (Lab)]

My Lords, I am grateful to the Minister for repeating the Statement made elsewhere. He was present for part of the debate on artificial intelligence on Monday. On reflection, it is a bit surprising that the Government were not able to accelerate the announcement of this new body. It would have helped a lot in that debate. No doubt the tyranny of the grid is to blame again, but many of us would have felt the benefit had we known, not least, that the membership of the board had been enhanced by those Members of your Lordships’ House already referred to.

To go back in history a bit, the Centre for Data Ethics and Innovation came out of amendments we proposed during the passage of the Data Protection Bill, but it was built on excellent work by the Royal Society and others. We should pay tribute to the groundwork that led to today’s announcement. Those amendments had a lot of support from around the House and would have gone into the Bill had we been able to push them further, but we could not get them within the bounds of the Bill’s framing. We should say clearly that the model we had in mind then was the independent Human Fertilisation and Embryology Authority. In preparing the thinking in this new area of advanced technology and data processing and protection, one needed a carefully balanced body that could regulate in the context of difficult ethical issues raised by research and development.

I will now ask a number of questions about the body itself, and I hope that the Minister will respond, in writing later if not now. The body was originally intended to be an independent statutory body, but it is not because no powers have yet been established. What is the progress on that? The reports I have read suggest that that is still an objective of the Government, although they are making a virtue of the fact that it is an advisory committee in the interim period. In some senses, they will probably be judging its success, which is a bit worrying given that the whole benefit would be that it was independent of government, long-term and able to look without fear or favour at the big issues. If it is an advisory committee of the department, how independent will it be in practice? Is funding secured? Can it spend what it needs to get the research and advice it needs? How much of the original thinking about the HFEA remains? As an advisory committee, can it request information? One problem is the difficulty of extracting information from the behemoths that populate the international information society.

The press release rightly describes the membership as “stellar”. Given the names already mentioned here, I think we should recognise that. I confess that my application was weeded out very early in the game. This was unfortunate, because I would have been delighted to be part of that. Having seen the full list and heard why they were chosen, it is clear that the right decisions have been reached and I bear no malice to those responsible—honest. If the membership question comes up later, I am still around.

In the absence of the new centre starting up, we have only two or three areas of activity. We have a statement as a result of the consultations that took place. It talks about the focuses being to provide clear guidance and regulation and to lead debate about how data can be used in the future. But there are still some problems that need to be resolved, and I will be interested to hear the Minister’s comments. The AI report we discussed at length in a very good debate on Monday, when there were notable speeches from the right reverend Prelate the Bishop of Oxford and the noble Lords, Lord Reid and Lord Browne, shows the range of issues that are going to be up for discussion. These are very abstruse areas of intellectual activity such as ethics and the nature of machines—whether they are responsible for their actions and, if so, how any redress can be obtained. The noble Lord, Lord Browne, posed questions about intelligent weapons and what controls must be placed on them. It is a very stretching agenda. All we know is that issues currently in the list include data trusts, algorithms and consumer experiences. I do not think there will be a shortage of those. Can the Minister explain what the process will be? I gather an overall strategy document will be revealed.

There are some concerns about the balance between advice and regulatory action. I think the plan would be for advice to be offered to government and regulatory action to be taken by existing or other bodies. Could we have confirmation of that? There is a question about the balance between ethics and innovation. Clearly, innovations are difficult to support if they raise big ethical issues too quickly; they often need to be tested over time and analysed. It would be useful if there were a way forward on that. Of course, there is the whole question of how the Government intend to treat public data, its use and value for money, and the extent to which it will be available.

Lastly, the new centre, which I wish extremely well, enters a rather crowded space with the Information Commissioner’s Officer, Ofcom and the CMA, all of which have statutory functions in this area, but perhaps I may counsel that also to come are the Alan Turing Institute, which is now up and running, and the Open Data Institute. Therefore, there will be a need for some time for this whole process to settle down and for leadership from the Government on how it will work.

The responses to the consultation showed a clear public wish for consistency and coherence, and I hope that in that process there will be room for consultation. I do not wish the new body to be a proselytiser for data or indeed for artificial intelligence, but there is a difference between proselytising and being in an explanatory mode, reassuring people and explaining to them the benefits as well as the risks of this new technology. The centre needs to be public facing and fully engaged in that process, and I wish it well.

[Lord Clement-Jones (LD)]

My Lords, I too thank the Minister for repeating the Statement. He was missed in the debate on Monday. I have had the benefit of reading the Government’s response to the consultation on the Centre for Data Ethics and Innovation. I share the enthusiasm for the centre’s creation, as did the Select Committee, and, now, for the clarification of the centre’s role, which will be very important in ensuring public trust in artificial intelligence. I am also enthusiastic about the appointments—described, as the noble Lord, Lord Stevenson, said, as “stellar” in the Government’s own press release. In particular, I congratulate Members of this House and especially the noble Baroness, Lady Rock, and the right reverend Prelate the Bishop of Oxford, who contributed so much to our AI Select Committee. I am sure that both will keep the flame of our conclusions alive. I am delighted that we will also see a full strategy for the centre emerging early next year.

I too have a few questions for the Minister and I suspect that, in view of the number asked by me and by the noble Lord, Lord Stevenson, he will much prefer to write. Essentially, many of them relate to the relations between the very crowded landscape of regulatory bodies and the government departments involved.

Of course¸ the centre is an interim body. It will eventually be statutory but, as an independent body, where will the accountability lie? To which government department or body will it be accountable? Will it produce its own ethics framework for adoption across a wide range of sectors? Will it advocate such a framework internationally, and through what channels and institutions? Who will advise the Department of Health and Social Care and the NHS on the use of health data in AI applications? Will it be the centre or the ICO, or indeed both? Will the study of bias, which has been announced by the centre, explore the development of audit mechanisms to identify and minimise bias in algorithms?

How will the centre carry out its function of advising the private sector on best practice, such as ethics codes and advisory boards? What links will there be with the Competition and Markets Authority over the question of data monopolies, which I know the Government and the CMA are both conscious of? In their consideration of data trust, will the government Office for Artificial Intelligence, which I see will be the responsible body, also look at the benefits of and incentives for hubs of all things? These are beginning to emerge as a very important way of protecting private data.

What links will there be with other government departments in giving advice on the application of AI and the use of datasets? The noble Lord, Lord Stevenson, referred to lethal autonomous weapons, which emerged as a major issue in our debate on Monday. What kind of regular contact will there be with government departments—in particular, with the Ministry of Defence? One of the big concerns of the Select Committee was: what formal mechanisms for co-ordinating policy and action between the Office for Artificial Intelligence, the AI Council, the Centre for Data Ethics and Innovation and the ICO will there be? That needs to be resolved.

Finally, the centre will have a major role in all the above in its new studies of bias and micro-targeting, and therefore the big question is: will it be adequately resourced? What will its budget be? In the debate on Monday, I said that we need to ensure that we maintain the momentum in developing our national strategy, and this requires government to will the means.

[Lord Ashton of Hyde]

I am tempted to say that I will write, but I will try to answer some of the questions, and I will write regarding some that I do not get around to. I was in at the beginning of the debate on AI and I listened to the noble Lord’s speech.

[Lord Clement-Jones]

My Lords, that is all that he needed to listen to.

[Lord Ashton of Hyde]

Not everyone would agree with that, but I did indeed listen to it. I have read that AI is a joint responsibility with BEIS, and my noble friend Lord Henley coped more than adequately, so I do not think that I really was missed.

There was a great deal of support for this innovation—the centre—both in the response to the consultation and, as the noble Lord, Lord Stevenson, said, in proceedings on the then Data Protection Bill, so I am grateful for that today, but I accept the very reasonable questions. On the centre’s independence as it stands now and its statutory establishment, I say that we have deliberately set this up as an advisory body so that it can consider some of the difficult issues that noble Lords have raised. Policy is the Government’s responsibility, so there should not be any confusion about who is held accountable for policy—and it is not the Centre for Data Ethics and Innovation. When this has been established, when we have seen how it has worked and when we have addressed the questions of the crowded space that both noble Lords mentioned, it is our intention to put this on a statutory basis. Then we will see how it has worked in practice. When it comes to putting it on a statutory basis, I have no doubt that there will be lots of back and forth in Committee and things like that on the exact definitions and its exact role.

There are some differences from the Human Fertilisation and Embryology Authority, although of course that was a particularly successful body. One of the main differences was that a lot of those things were considered in advance of the science, if you like, and before the science was put into place. With AI, it is here and now and operating, so we do not have a chance to sit back, think about it in theory and then come up with legislation or regulation. We are dealing with a moving target, so we want to get things going.

As far as I am aware—I will check and write to the noble Lord, Lord Stevenson—the centre has no specific powers to demand information. That is, of course, something that we can look at when it comes to being on a statutory basis.

I am sorry that the application for membership by the noble Lord, Lord Stevenson, was not accepted. There can be only one reason: he spends so much time on the Front Bench that he would not have time, because we expect the directors to spend two to three days a month attending this, so it is a very large work commitment.

As noble Lords will know, the work plan includes two initial projects, which were announced in last year’s Budget: micro-targeting and algorithm bias. We expect the centre, in discussion with the Secretary of State, to come up with a work plan by spring 2019. As the noble Lord, Lord Stevenson, mentioned, there is a tension, if you like, between ethics and innovation, but we are very keen that it consider both because we have to be aware of the potential for innovation, which is constrained in some cases. We would not want a situation where the opportunities for AI for this country are avoided. As the report by the noble Lord, Lord Clement-Jones, made clear, there are tremendous opportunities in this sector. We are aware of the tension, but it is a good tension for the centre to consider.

Both noble Lords talked about the crowded space in this area. We expect the centre to produce memorandums of understanding to outline how it relates to bodies such as the AI Council, which has a slightly different focus and is more about implementation of the AI sector deal than considering the ethics of artificial intelligence. We understand that they need to work together and expect the centre to come back on that.

The noble Lord, Lord Clement-Jones, asked about accountability. The centre will be accountable to the Secretary of State for the DCMS. That is clear. He will agree its work plan. Of course, in terms of independence, once he has established that work plan, what the centre says will not be up to him, so there is independence there. We included in our response that the Government will be expected to reply within six months, so there is a time limit on that. It will apply to all government departments, not just the DCMS. The Ministry of Defence and the department of health have obvious issues and the centre can provide advice to them as well.

The noble Lord, Lord Clement-Jones, asked whether the centre, when it considers bias, would include audit mechanisms. It absolutely might. It is not really for us to say exactly what the centre will consider. In fact, that would be contrary to its independence, having been given the subject to think about. In our response we said some of the things that might be considered, such as audit mechanisms.

There is an obvious issue about competition, which the House of Lords Select Committee mentioned. Work is going on. The Chancellor commissioned the Furman review to look at that and we expect the centre to come up with a discussion on how it will work with the Competition and Markets Authority, but obviously competition is mainly to do with the Competitions and Markets Authority.

At the moment, the body is resourced by the DCMS. In the 2017 Budget, it was provided with £9 million in funding over three years. We expect that to be sufficient but, clearly, we will have to provide adequate resources to do an adequate job.