On Humane Tech: Facial Recognition, the Russian-Ukraine War, and the Future of Surveillance
On Humane Tech: Facial Recognition, the Russian-Ukraine War, and the Future of Surveillance with guest Katina Michael
The Lincoln Center for Applied Ethics hosts a weekly conversation “On Humane Tech,” highlighting relevant news in a conversational format with our team. Each week the topic changes, but one thing stays the same — we want to hear from you. Respond to our conversation below.
This week’s topic: Facial Recognition, the Russian-Ukraine War, and the Future of Surveillance with guest Katina Michael.
We continue our conversations about technology surrounding the Russian-Ukraine war with Katina Michael, professor in the School for the Future of Innovation in Society and School of Computing and Augmented Intelligence at Arizona State University.
Erica O’Neil, Research and Project Manager: This is a series of informational conversations with experts in areas of different tech as it relates to the Ukraine-Russia war. And we were really interested in getting your perspective today on the positives and negatives of Clearview AI and facial recognition technology capabilities. There was a recent article published in the New York Times detailing how Clearview made its AI available to the Ukrainian Government free of charge in order to identify Ukrainians and Russians in various contexts, so that they can kind of say who they are, and it has a lot of implications for privacy and dissent in a war zone or dissent in Russia and it’s been used across all this for different contexts. So we are just curious on picking your brain and hearing what you see to be the pros and cons of using something like this in such a heated environment.
Katina Michael: So the first thing to say is that all technology is inherently political. This particular story is about weaponizing facial recognition systems. Whereas once our face was used for identification purposes to afford us access to services or repatriate displaced persons, the technology has also been used to detect spies, dissidents, protestors, and activists for state surveillance. The adoption of new technologies in new contexts such as war, tests what might work on a mass scale and what does not, what citizenry will accept and what they will not. In other words, what are the limits of facial recognition technology in a public setting where the “system” is purportedly “open”, and in the name of “national security” and “social securitization”?
This is also very much about the military-industrial complex. We’ve seen a number of private organizations whose business it is to build artifacts, software and hardware and corresponding components, seek to conduct evaluative demos of their facial recognition technology to various government agencies and policing organizations, touting “tech for good” or “beneficial tech”, and its potential in asymmetric conflicts such as war zones. More cameras in smart cities mean more multimedia data collection; images that can be captured and parsed using facial recognition and other video analytical techniques, near real-time. This is a new breed of “just-in-time” tech, that is in the field “overt” in some ways that humans might see the embedded tech in lampposts, and “covert” in other ways as the analysis takes place “invisibly” and “after the fact”. Into the future, emotion detection systems may well be pre-emptive in warning of impending civil disobedience or even war or even soldiers that are mentally or physically fatigued.
Victoria Vandekop, Communications Program Coordinator: How does facial recognition technology work, especially when it’s used in such different contexts?
Michael: Well, one way to test facial recognition technology “at scale” is to position it as either a perceived weapon of defense for the “common good”, or even as a compassionate technology that will aid, for example, in the identification of fallen soldiers in enemy territory. The Ukraine and Russia war involves soldiers and citizens who are predominantly white. Facial recognition technology performs well in contexts where white males are heavily represented, although we do not know the technology’s accuracy in “noisy” environments in the field where one-to-many (1:N) matches are presupposed, or where there is a greater diversity in race, gender, and ethnicity. The question is, how does facial recognition technology stack up in unstructured environments, with possible poor lighting (especially at night or bright sun), various climate exposures (rain, snow), tilting of heads that are not still, apparel like helmets, scarves or balaclavas and other disruptive environmental conditions?
So if you have a passport-style still shot registered in a database, and you personally present in front of a controlled camera setting to verify who you are, you’re trying to do what is called a one-to-one match. That’s when photographs have been officially registered just like when an individual unlocks their smartphone using their facial image. In the case of soldiers or people in the field who are roaming and unaware their facial image is being captured and compared to a large database of faces that have been scraped from the Internet (and whose personal identity is also unknown) that can prove to be a lot harder and fraught with significant issues sometimes categorized as false positives or false negatives. One search can return many matches and the accuracy of a match in this instance is highly debatable and based on what is called “degrees of confidence”. Clearview AI has amassed a facial image dataset of 10 billion images, by web scraping open source information (OSI), from public social media profiles among many other online sources. In this instance, people are not really cooperating when they’re in the field walking around minding their own business, completely unaware their face is being photographed and stored in the cloud.
Vandekop: Considering the use of facial recognition in war, what does all of this mean for the general public?
Michael: I think what we’re looking at here has more to do with “blanket coverage” surveillance of the populace; and less about identifying fallen soldiers or enemy spies in one another’s territory. Sure, purportedly drones can conduct signature strikes at a micro level using a “kill list” relying possibly on facial image matches in the future, but is this war now being used as a front to test the latest “technology for good” scenario? Prior to the Aadhaar system in India, the largest biometric databases were not more than 50 million records in size. In what other context would this kind of facial recognition technology be considered proportional or even viable for the individual citizen? Imagine, the whole world had a grid of internet-worked smart cameras and every single person on earth had a registered image in a global database for emergency purposes. But who is the enemy? And I won’t get into the geopolitics here because that’s besides the point, the point is that we’ve got commercial technology which in 1993 through the joint effort of the Defense Advanced Research Project Agency (DARPA) and the Army Research Laboratory (ARL) established the face recognition technology program FERET to develop automated face recognition capabilities for security and intelligence and law enforcement. It doesn’t take any stretch of the imagination to see how the technology might be used in the future.
Check in with us next week for the second part of our conversation with Katina Michael.
Original Source: “On Humane Tech: Facial Recognition, the Russian-Ukraine War, and the Future of Surveillance”, Medium, https://medium.com/@lincolncenterappliedethics/on-humane-tech-facial-recognition-the-russian-ukraine-war-and-the-future-of-surveillance-with-a26168bb64c4