Enlisting AI in our war on coronavirus: Potential and pitfalls
Given the outsized hold Artificial Intelligence (AI) technology has acquired on public imagination of late, it comes as no surprise that many are wondering what AI can do for the public health crisis wrought by the COVID-19 coronavirus.
A casual search of AI and COVID-19 already returns a plethora of news stories, many of them speculative. While AI technology is not ready to help with the magical discovery of a new vaccine, there are important ways it can assist in this fight.
Controlling epidemics is, in large part, based on laborious contact tracing and using that information to predict the spread. We live in a time in which we constantly leave digital footprints through our daily life and interactions. These massive troves of data can be analyzed with AI technologies for detection, contact tracing and to find infection clusters, spread patterns and identify high-risk patients.
There is some evidence that AI techniques analyzing news feeds and social media data were not too far behind humans in originally detecting the COVID-19 outbreak in Wuhan. China seems not only to have used existing digital traces but also enforced additional ones; for example, citizens in Nanjing are required to register their presence in subway trains and many shops by scanning QR codes with their cellphones. Singapore, lauded for its effective containment of the virus without widespread lockdowns, has used public cameras to trace the interaction patterns of the infected, and even introduced a crowd-sourced app for voluntary contact tracing. In the U.S., research efforts are underway to mine body temperature and heart-rate data from wearables for early detection of COVID-19 infection.
It is widely feared that COVID-19 cases, at their peak, will overwhelm medical infrastructure in many cities. Evidence from Hubei, China, and Lombardy, Italy, do indeed support this fear. One way to alleviate this situation is to adopt novel methods of remotely providing medical help. The basic infrastructure for telemedicine has existed for a long time but has been a hard-sell from consumer, provider and regulatory points of view — until now. Already, the U.S. has waived regulations to allow doctors to practice across state boundaries; the Department of Health and Human Services (HHS) has also announced that it will not levy penalties on medical providers using certain virtual communication tools, such as Skype and FaceTime, to connect with patients.
AI technologies certainly can help as a force-multiplier here, as front-line medical decision support tools for patient-provider matching, triage and even in faster diagnosis. For example, the Chinese company Alibaba claims rapid diagnostic image analytics for chest CT scans; China also has leveraged robots in disinfection of public spaces. Remote tele-presence robots increasingly could be leveraged to bring virtual movement and solace to people in forced medical quarantines.
AI technologies already have been enablers of, and defenders against, fake news. In the context of this pandemic, our incomplete knowledge coupled with angst has led to an infodemic of unreliable/fake information about coping with the outbreak, often spread by well-meaning (if gullible) people. AI technologies certainly can be of help here, both in flagging stories of questionable lineage and pointing to more trusted information sources.
AI also can be used to distill COVID-19-related information. A prominent example here is a White House Office of Science and Technology Policy-supported effort to use natural language-processing technologies to mine the stream of research papers relevant to the COVID-19 virus, with the aim of helping scientists quickly gain insight and spot trends within the research. There is some hope that such distillation can help in vaccine discovery efforts, too.
Suppression by social distancing has emerged as the most promising way to stem the tide of infection. It is clear, however, that social distancing — like sticking to a healthy diet — runs very much counter to our natural impulses. Short of draconian state enforcement, what can we do to increase the chances that people follow the best practices? One way AI can help here is via micro-targeted “behavioral nudges.” Like it or not, AI technologies already harvest vast troves of user profiles via our digital footprints and “weaponize” those for targeted ads. The same technologies can be readily rejiggered for subliminal micro-targeted social distancing messages that could include distracting us from cabin fever. There already is some evidence that mild “nudging” can even reduce the sharing of misinformation.
Lockdowns and social distancing measures are affecting the education of millions of schoolchildren. Tutoring services assisted by AI technologies can help significantly when students are stuck at home. China reportedly has relied on the help of online AI-based tutoring companies such as Squirrel AI to engage some of its millions of schoolchildren in lockdown.
And while self-driving cars remain a distant dream, delivering essential goods to people via deserted streets certainly could be within reach. Depending on how long shelter-at-home continues, we might rely increasingly on such technologies to transport critical personnel and goods.
Some of these potential uses of AI are controversial, as they infringe on privacy and civil liberties or reflect the very type of applications that the AI ethics community has resisted. Do we really want our personal AI assistants to start “nudging” us subliminally? Should we support increased cellphone tracking for infection control? It will be interesting to see to what extent society is willing to adopt them.
Indeed, our readiness to try almost anything to fight this unprecedented viral war is opening an inadvertent window into how we might handle the worries surrounding an AI-enabled future. Ideas such as universal basic income (UBI) in the presence of widespread technological unemployment, or concerns about diminished privacy thanks to widespread AI-based surveillance — all are coming to the fore.
China has mobilized state resources to feed its quarantined population and used extensive cellphone tracking to analyze the spread of the virus. Israel is reportedly using cellphone tracking to ensure quarantines, as did Taiwan. The U.S. is considering UBI-like ideas — e.g., providing thousand-dollar checks to many adults effectively unemployed during the pandemic — and is reportedly mulling cellphone-based tracking to get people to follow social distancing guidelines.
Once such practices are adopted, they will no longer just be theoretical constructs. Some or all of them will become part of our society beyond this war on the virus, just as many Great Depression-era programs became part of our social fabric. The possibility that our choices in this time of crisis can change our society in crucial ways is raising alarms and calls for circumspection. Yet, to what extent civil society is likely to pause for circumspection at the height of this “execution imperative” remains to be seen.
Subbarao Kambhampati, PhD, is a professor of computer science at Arizona State University and chief AI officer for AI Foundation, which focuses on the responsible development of AI technologies. He served as president and is now past-president of the Association for the Advancement of Artificial Intelligence and was a founding board member of Partnership on AI. He can be followed on Twitter @rao2z.
The Hill has removed its comment section, as there are many other forums for readers to participate in the conversation. We invite you to join the discussion on Facebook and Twitter.