Power with responsibility: deploying AI in the struggle against COVID-19
Governments are striving to beat coronavirus, often turning to emerging technologies such as AI to help them do it. These offer tremendous benefits in times of crisis and otherwise. But as Dr Steve Bennett explains, such technologies must be used carefully if their full potential is to be reached – and the risks avoided.
“Nothing beats a crisis to push governments to do things faster than they might normally want to do,” says Dr Steve Bennett, director, global government practice at analytics software specialist, SAS. COVID-19 is a prime example. Since the outbreak, governments around the world have had to accelerate digital transformation programmes, in some cases delivering projects in months or even weeks rather than years.
This is driving demand for analytics tools that can help in the fight against the virus, in shoring up economies, and in maintaining and scaling up existing public services. Bennett gives the examples of working out how best to roll-out trillions of dollars of stimulus packages; optimising intensive care beds; and running epidemiological models to inform policy decisions about when and how to go about reopening society – all things that have to be informed by good science and good data.
Artificial intelligence (AI) plays no small part in this. “When we think about the phases of a pandemic, AI can be helpful across the spectrum,” says Bennett, who has a PhD in computational biology and spent 12 years at the US Department of Homeland Security before joining SAS. “Firstly, it can be used even before a pandemic – where do you see humans interacting with animal populations or encroaching on new ecosystems because of population growth or climate change? You can use AI to identify hotspots where an outbreak might occur.”
Once a virus is spreading, AI and machine learning can be used to detect it quickly – for example by scouring Twitter for evidence of health events and geolocating data to discover trends, Bennett explains, while on the treatment side, it can be used to analyse chest x-rays to help doctors make faster diagnoses. And then, once a pandemic is over, Bennett adds, it can be used to test different policy scenarios, helping those in government to determine which mitigation strategy is likely to have the most positive impact in tackling a future health crisis. “From detection to response and hopefully also prediction, prevention and preparation, the value of AI is huge,” says Bennett.
Safeguarding citizens
The need to get to grips with issues that are seldom tackled at such scale and speed is propelling innovation both within government and in the private sector companies they buy from. But some of the emerging technologies now being embraced as a result aren’t yet covered by regulation.
We’ve all heard reports about AI systems turning out biased outcomes because the algorithms behind them have been inadvertently trained with biased data sets.
“In government, the stakes are high because agencies may be using AI to make decisions that affect citizens significantly,” Bennett says. “It could deny a benefits claim, delay a tax refund or, in the extreme, it could be used in the policing context as the basis for detaining someone and depriving them of their liberty.”
The biases that lead to such decisions could have a particularly detrimental effect during the current pandemic, potentially making situations worse for people already at a disadvantage. “There’s always the risk, when we’re moving faster than usual, that things fall by the wayside,” Bennett says. “The less fortunate in our communities are suffering differently from COVID than other sections of those communities, so now more than ever thinking about how to avoid bias upfront before using these tools is really important.”
“Yes, it’s great to have the technology in place but it’s tough to build things into the aircraft while you’re flying it in a crisis like this,” he adds. “You need to step back and, at the policy level, have a discussion about the values in society that you want to make sure are reflected in the technology.”
One recommendation SAS makes is that no decision that could have a negative impact on a citizen should ever be made solely from the output of an algorithm. “There should always be a human in the loop,” Bennett says.
National AI strategies
Progress is being made in this area. Many countries are working up or have implemented national AI strategies including Canada and the US, while the United Arab Emirates has a dedicated AI minister, the first country in the world to do so. Most strategies, Bennett points out, have two components: encouraging the adoption of AI by government and the private sector; and working out how to take advantage of the benefits it offers in a way which protects societies’ values.
Of course, those societal values differ from country to country, and those differences have implications. “Countries that have strict privacy protections and imposed limits on what they can do with AI and government data are worried about falling behind in the research race,” Bennett says. And there are more serious concerns. Bennett cites Vladimir Putin, who once said that whoever becomes the leader in the sphere of AI “will become the ruler of the world”. “There are a lot of political leaders talking about AI as a potential way to have strategic advantage over their adversaries,” Bennett says.
To counter this, he advocates collaboration. “Countries working together is always better than not. That doesn’t mean they have to give away their intellectual property, but these technologies can be used for good and they can be used for bad and there’s a discussion to be had about the benefits and the risks.”
The privacy debate
More broadly, when it comes to analytics tools and the data needed to feed them, privacy issues often come to the fore. COVID-19 contact-tracing apps, many of which are powered by AI technology, are being used by several countries around the world including Germany, India, Singapore, Norway, and Australia. In each case, tech experts and civil liberties groups have raised concerns about privacy and how the data collected by such apps is used. “We should always be suspicious when the government wants to collect data about us – that’s healthy,” Bennett says. “And in Western governments, protecting privacy tends to be a key concern.” But he points out: “When we look at privacy regulations around the world, most of them include a carve-out for exigent circumstances, such as public health emergencies – so it is possible to use data legally within those frameworks in crises like this.”
Bennett is sceptical about the effectiveness of contact-tracing apps, primarily due to compliance issues and his concern that they flood public health authorities with alerts about individuals who are unlikely to have contracted the virus. Data collected by apps can be useful, Bennett says, but when used as supporting data, not the primary source.
The approach SAS has taken to contact tracing is to adapt its existing Visual Investigator product. Developed originally for use by agencies responsible for law enforcement, national security, and child welfare, the company has customised it so that it can be used to search for patterns in large networks of people at speed. Rather than take over the role of a human contact tracer, it effectively enables their work to be done more quickly.
Free people up to do the thinking that only humans can do
This focus on speeding up processes is key, in time of crisis and otherwise. Bennett says there are numerous examples of AI having “hugely positive impacts” in government, improving efficiency and effectiveness in anything from tax collection to citizen services.
He understands agencies’ concerns that using such technologies may put people out of work but insists this won’t be the case. “What they’re going to do is to automate manual tasks, freeing people up to do the thinking that only humans can do. I’m very optimistic about these tools being used in such a way as to put civil servants around the world to better use in advancing solutions to problems and to help make life better for citizens.”
Fundamentally, this ties in to decision-making. “The goal in all that we do in data science, analytics and AI at SAS is to help people in government make better decisions faster and more cheaply,” says Bennett.
If the risks can be mitigated, the potential rewards are significant, but there remain barriers in government: inadequate data sharing between departments is one; lack of cultural willingness another. “You could have the best technologies in the world but agencies have to be willing to use them. There’s a whole conversation around the cultural willingness to adopt these kinds of technologies,” Bennett says.
“In the short term, I think governments are going to continue to use AI for limited use cases – for speeding up fraud detection and getting benefits claims out the door more quickly,” he says. “Over the mid- to long-term we’re going to see these technologies handle the more strategic policy questions.”
The use of AI in the era of coronavirus is putting the technology centre stage, accelerating its use by governments with huge scope for improved performance. As any tech expert will tell you, AI can be used for good and for evil but put the proper safeguards in place early, and there are an almost-endless myriad of benefits we are only just beginning to grasp.
Join SAS for a complimentary virtual event. What’s ahead for analysing data at the edge, in the cloud, and with AI and partnerships? SAS executives will talk about how innovation is the key to thriving in uncertain times and beyond. Simon Sinek’s featured keynote discussing The Infinite Game will be available through July 16.
In the government executive connection “Leading the Way with Analytics in COVID-19 Recovery” hear how public sector agencies, at all levels, are supporting individuals, families and communities in recovery efforts and why analytics continues to play a critical role in these missions. Register here for on-demand viewing.
About Partner Content
This content is brought to you by a Global Government Forum, Knowledge Partner.
Related Posts
