On Monday 8 June, we were delighted to kick off CogX 2020 by curating the first day of the Ethics & Society stage. 25 speakers across six panels joined us to tackle the knotty, real-life trade-offs of benefits and harms that emerging technologies bring to people and society. This is a summary of the first panel of the day – Does Whitehall need more WEIRDos? – which you can watch in full below:
John NaughtonSenior Research Fellow, CRASSH, University of Cambridge
Jeni TennisonVice President and Chief Strategy Adviser, Open Data Institute
Chi OnwurahLabour MP for Newcastle upon Tyne Central, Shadow Minister for Digital, Science and Technology
Jack StilgoeAssociate Professor of Science and Technology Studies, UCL
The slogan of Silicon Valley, introduced by a young Mark Zuckerberg, was ‘to move fast and break things.’ Slogans are fine for campaigning, but no use in governing. This session is about the effective use of data in governing, so let’s get started.
Today I’ll be walking the five miles to Parliament later, following our Government’s decision to abandon the digital parliament that enabled MPs to contribute while following Government guidance to work from home. This has been a clear example of how the issue there was not the technology, it was the people. Which brings me to Dominic Cummings.
To link to the topic of this session, Cumming’s claim that we need more weirdos and misfits in Whitehall is one I completely reject. All he is really saying is ‘we need more people like me’. It reflects an age-old civil servant fear of opening up the Civil Service, and roles in government, to true meritocracy. Cummings sees himself as a misfit, but he perhaps also doesn’t recognise that he is also white, male and privileged. His recruitment strategy would ensure more people like him.
I also reject the notion that this is just about government data. We also have less trust in the use of data across public and private sectors. From Google and Facebook hoarding data, to Amazon and Netflix telling us what to buy and watch, data use is everywhere, and trust is diminishing. But there is no meaningful difference in law for different uses – whether it’s cancer research or big tech, uses of data all abide broadly by the same set of laws. And that, coupled with the erosion of trust, is a significant issue.
The challenge here is that Government is unique. It’s a monopoly provider of services and serves everyone, including the most vulnerable. To build trust in public sector data use the Government must recognise that technology must empower people, not control them. Data sharing can improve society and inform and improve public services, making them more efficient and more versatile. But data sharing with private firms undermines this trust because the accountability and transparency isn’t there.
The best way to empower citizens and build trust is to have a government which recognises citizen’s digital rights and gives them more control. This is contingent on the transparent and effective data infrastructure but – and the COVID app is a good example – this infrastructure tends to be centralised. This opens up risks for hacking and breaches, but also creates fears and uncertainty about who else is getting access to that data about us.
The only point I agree with Dominic Cummings on is that more diversity is needed in Whitehall – diversity in talent, in thought and in people. If we look at the anger witnessed in the UK in response to the killing of George Floyd, it is because of the lack of change in diversity in Whitehall. In addition, Whitehall tends to prefer generalists, not specialists. So, as well as diversity in people, we need more experts in data management and infrastructure. An example of how COVID data is collected is an example of that lack of expertise.
We need more Whitehall experts who recognise and believe in empowering, not controlling people with data. People should have agency and control over their own data. More of the weirdos that Cummings advocates for will not bring that.
I agree that diversity in government is vital. If we were to be generous to Cummings’ job ad, we’d say ‘yes, you’re right, we do need more diversity in Whitehall’. But that’s not what he’s doing; he’s saying we need more thinkers, but they have to have read the same five AI papers that I’ve just read on Nature. That’s not diversity.
Expertise in Government is peculiar because it is hugely visible right now – because of things like SAGE (the Scientific Advisory Group for Emergencies). But we kid ourselves with a Hollywood view of expertise, where we can call on the expert to swoosh in from the lab and give us an answer. Experts don’t do that. They come in and provide evidence to help navigate uncertainty, not just give all the answers. To navigate that uncertainty, we need cognitive diversity and to avoid groupthink.
Coming out the other side of COVID-19, we’ll see how the dangers of groupthink have been at play. For me, it’s not: what do we know or not know about COVID-19, it’s what did we learn? Because this has been a failure of learning. In the case of COVID, we knew the virus was coming but we didn’t act. This isn’t a failure of science, it’s a failure of Government to learn.
The current fixation on data science is part of a broader historic fixation and fascination with all scientific fields. And within that, data maximalism is a real challenge. There’s a quote from Cambridge Analytica that ‘there’s no data like more data’. There is that urge to throw more data at the problem, but data has its own politics in who gets represented, who gets left out, who gets burdened and who benefits from it. So, the challenge isn’t just to get data scientists into the mass of humanities graduates, it’s to get some humanities graduates into the mass of data science.
My final point is around the ‘crystal ball’ vision of data science to predict the future, these ‘super–forecasters’ who Cummings covets. The question isn’t ‘can we gather information about the future better?’ The question is an ancient question of ‘even if we did know the future, would we be able to persuade people to act upon it?’ If we take something like climate change, it’s not the knowing that’s the problem, it’s the changing behaviour, it’s acting on it.
Diversity in decision making is good but let’s also add a further challenge and not pretend dotting in a few clever people into Government is going to solve our problem. We have had decades of underinvestment in order to understand things in scientific terms. There is a serious problem with where Government is getting its science from, it’s no longer able to fund that science itself but we haven’t yet come up with a good alternative.
I also assume the misfits and weirdos referred to are data scientists, physics and maths grads – people who can build models which might help us understand what the future may or may not hold. And my reaction is that there are some things which are compelling with that, and some things to be concerned about.
What’s compelling is that we do need data, evidence, research and science at the heart of government. It helps us understand economy and society, which is especially at times like these. Also evaluating the impact of polices and interventions – like the impact and effectiveness of the contact tracing app – will help us whether to invest in it more or whether to try a different approach. And data modelling helps us understand the potential impact of certain interventions before we do them. That saves time and can also help us avoid taking the wrong path.
But it’s not like there aren’t those skills already. There’s the Government Statistical Service and data science teams. This isn’t a new team needed; rather we need to support the teams that are there already. Having these skills is important because data, digital and tech impacts on us all. Understanding how data sharing from government can help research and innovation is an important aspect of policymaking.
But there are concerns. I worry that those with strong maths, physics, or computer science backgrounds are trained to see data as abstract and numbers as something you just play with. That’s fine in a physics model, but when we talk about policy those numbers represent and impact on people’s lives. This means they have a different weight and quality to them. And the data and numbers they’re playing with will be biased in the way it is collected due to structural racism and the society we have dictates the data we have about it – what we choose to collect.
There’s also a reductive aspect – reducing people’s experience to numbers diminishes us and our ability to understand the deep impact of policy. Reducing our lives to numbers becomes utilitarian – we just measure the numbers, not the rich improvement in quality of life that we should aim for.
And more data isn’t necessarily better when data collection and the feeling of surveillance impacts on our privacy and autonomy. And so, we need not just data scientists, we also need social scientists that understand how biases can affect what we conclude and so on.
The other catch around introducing misfits and weirdos is that embedding change across Whitehall and the public sector is a huge task. It requires soft skills – the translation of findings from science and data into policy and to understand that they’re there to inform, not dictate policy. Introducing this across the public sector is a broad transformation task and you don’t get that from a small central cadre of centralised people trying to make that change happen. You need those skills across the Civil Service and to also reach beyond and outside government to get support from everyone else.
I support more data, science and evidence in Government, but we also need the complementary skills and real diversity to see beyond the numbers to make a positive difference to people’s lives.
I’m reminded of Barack Obama’s first chief of staff’s statement: ‘Never let a good crisis go to waste.’ And, in this context, the COVID-19 crisis is such a useful crisis in some respects. Particularly, for example the concerns and disputes about the contact tracing apps have highlighted a whole host of challenges, like whether to make the data centralised or not. This raised the question as to whether the purpose of technology is to increase human agency or decrease it. We realise the centralised / decentralised debate is at its heart that same philosophical debate.
I’m also reminded of an early feminist slogan: ‘the personal is political’. Similarly, the technological is now political. And so I wonder with these assumptions about data science, that if we have more data everything is better. I simply don’t believe that. Can more ever equal better?
Philosophers of science have said about the epistemological fear of data science: that knowledge falls out of data and so data will replace knowledge. But this obviously isn’t true. The worry with crises creating opportunity is that a crisis is more easily capitalised on by the privileged and powerful. With the contact tracing app, there are problems of function creep here. We might be more comfortable with more surveillance in extreme circumstances, but many will want to capitalise on that. So we need to ensure that those who wish to gather more advertising data on us or spy on us, for example, can’t continue to do so or do so more effectively because of this pandemic.
The aim with a centralised approach to the contact tracing app is not just to improve contact tracing but to improve understanding of the disease. Centralised databases enable better research. But the problem is that so far not enough data is shared with those who can act on it. The empowerment of local authorities, charities and communities is dependent upon centralised data being shared with them, so that the insights from that data and the agency to act on it are shared equally with everyone. If that access isn’t granted, I worry that centralisation disempowers those communities and individuals.
The reason we don’t have a viable app now is because we went down the centralised route. If we had decentralised we might have an app now which could get us out of lockdown. There is this dream that if only we could collect all the data in the world we could control the world. But we won’t control the world because that data is just a digital twin which will be biased and incomplete.
COVID has meant we’ve dashed online in a hurry. We should build on this to embed the right principles of data empowerment, but I fear it’s the power grab by certain tech companies and those in government that will be the legacy of this.
There’s a techno–solutionism here: an assumption that technology will do better than people. But manual contact tracing has a long history. There’s a small authority in Wales who have deployed manual tracing very effectively. This crisis has revealed the hyper centralisation of the British state. A central bureaucracy which has assumed it is the only one with the resources to act on this. Why’s no one talking about this?
They do in the regions, but Whitehall drives the funding, the decisions and the ideology that government should get out of the way of the economy. This has minimised local authorities’ agency.
I’m going to interject on that final big question as I don’t think we will have time to answer it. Thank you to all the panellists for a fascinating conversation which surfaced serval issues which we will return to throughout the day.
Report with recommendations and findings of a public deliberation on biometrics technology, policy and governance
Findings from a rapid expert deliberation to consider the risks and benefits of the potential roll-out of digital vaccine passports
A research partnership with NHS AI Lab exploring the potential for algorithmic impact assessments in an AI imaging case study
The societal impacts of introducing a public health identity system: legal, social and ethical issues
The second in our series of events addressing the nascent ‘public health identity’ systems developing around the world.