UK’s response to COVID-19 mostly reliant on traditional data analysis, not AI

Beginning in March 2020, the government's Centre for Data Ethics and Innovation (CDEI) has been monitoring examples of how data-driven technology has been used in response to the COVID-19 pandemic - documenting them within what it calls a ‘COVID-19 repository'. From high profile applications such as contact tracing, to lesser known use cases that include the use of drones to deliver medical supplies to remote locations, data has been critical in tackling the virus.

However, despite artificial intelligence (AI) grabbing most of the headlines over the past year, and government ministers throwing the term around to gain kudos, it seems that it has been traditional data analysis and data sharing agreements that have been doing most of the grunt work. 

And somewhat surprisingly, the latest research from CDEI also states that there is broad public support for the use of data-driven technology in the fight against COVID-19 - in principle. This support even extends to newer technologies, such as the use of wearable devices to track social distancing in the workplace. 

However, the main blocker to this support is trust in the rules and regulations governing technology. 

Commenting on the findings, Edwina Dunn, Deputy Chair for the Centre for Data Ethics and Innovation, said: 

Data-driven technologies including AI have great potential for our economy and society. We need to ensure that the right governance regime is in place if we are to unlock the opportunities that these technologies present. The CDEI will be playing its part to ensure that the UK is developing governance approaches that the public can have confidence in.

Traditional tools still effective

The UK's response to COVID-19 has largely relied on existing data and analytics tools, which previously didn't have the political backing to be brought forward. The use of ‘shiny and new' AI has largely been used in healthcare settings and for vaccine research, which are clearly important, but it has not been as widely used as the public may be led to believe. 

As the CDEI notes, data has managed to provide some grounding of thought in a time of extreme uncertainty. Data has been used for tracking infection and death rates, as well as to monitor population movements through the use of geolocation and telecoms data. Meanwhile hospital asset data has been analysed to identify where equipment is free and where it is needed. 

The report notes: 

In contrast, AI and machine learning take-up was minimal. While the repository did capture some use-cases where AI was being deployed, we did not see evidence of widespread adoption and use-cases were mostly clustered in the healthcare setting.

In addition to this, the data being used at the heart of the pandemic response was by and large from existing datasets. Many found that additional data collection was not immediately necessary, instead augmenting and repurposing existing data for the specific COVID-19 use cases. For example, many organisations have made population and patient-level data publicly available to aid COVID-19 research and decision making. 

New data sharing agreements were also put in place, in a variety of forms. We have noted previously how some local authorities were ahead of central government in their work to share data and respond to the imminent public health crisis. 

The CDEI explains: 

New data sharing efforts have taken a variety of forms. In some cases, the government and public services have opened up datasets to the private sector for the first time (for example, by giving supermarkets access to information about vulnerable patients most in need of assistance). In other instances, we have seen individual public services pool their datasets, allowing for more sophisticated data analysis. This includes some children's services providers in London, which chose to collect and combine data on service performance in order to identify early signs of system stress.

While seemingly straightforward to administer, these data sharing initiatives required new legal agreements, oversight measures, technical standards and data storage tools - most of which had to be established in a matter of weeks of the first UK lockdown.

Public trust

As noted above, even though AI hasn't featured heavily in the UK's response to the COVID-19 pandemic, when looking at the use of data-driven technologies broadly, there is more public support than some may have previously thought. 

The CDEI commissioned Deltapoll to conduct a longitudinal survey of UK public opinion with a representative sample of over 12,000 individuals, running from June to December 2020. The results suggest significant public support for the use of data-driven technology over that period. Almost three quarters (72%) of the UK population felt that digital technology had the potential to be used in response to the outbreak - a sentiment shared by all demographic groups.

However, many respondents also felt that the potential of data-driven technology was not being fully realised. Fewer than half (42%) said digital technology was making the situation in the UK better (although only 7% claimed it was making matters worse). 

The CDEI suggests that this points to an "opportunity gap" - a chasm between technology's potential and the perceived reality of how it has been applied.

It goes on to argue that this gap is largely driven by a lack of trust in the governance of AI and data-driven tools. The report argues: 

The overall picture presented by our polling is one of a public that is largely sympathetic, and in some cases enthusiastic, about the idea of AI and data being used to tackle the pandemic. But that this has been in spite of, rather than because of, the way the technology has been deployed in different contexts. This suggests that public support is tenuous and dependent on trust in the governance of technology.

When asked the main reason why digital technology might not be effectively used in the COVID-19 response, a plurality of respondents cited concerns about whether people and organisations would be able to use the technology properly (39%). This was more than double the proportion of the public who pointed to problems with the technology itself (17%).

Some respondents also expressed misgivings about the governance of data-driven interventions. Whilst 43% of the public said existing rules and regulations were sufficient to ensure the technology is used responsibly, still close to a quarter (24%) disagreed. Trustworthy governance will enable the UK to make better use of data-driven technology and close this opportunity gap.

For organisations and policymakers looking to realise the benefits of greater data use, it will be important to build data governance mechanisms that are capable of building long term trust, and the CDEI is committed to playing its part.

My take

There's a few elements to this that are interesting. Firstly, the public appetite for using data to drive better public services is there, but that trust needs to be built. The government can't storm ahead and just assume the public will follow it. If mistakes are made down the line without the correct governance in place, and full transparency, that trust may be too hard to build back. The other point worth mentioning is that government institutions and organisations have felt that they can pursue innovative uses of data during this time - even if its using traditional tools - because there has been the political and public will. Those organisations need to act on this capital gained and work directly with the public on use cases that extend beyond COVID-19, whilst that appetite is still there.