BY PIETRO VIGILANZA
Since the start of the global pandemic, our dependence on technology has crystallized to an extent that would have seemed farfetched this time last year. While this certainly has some glad that we can function while we stay physically distanced to cut down on disease, it has also sparked several questions concerning the way we interact with the most recent innovations around us. For example: how do we keep our data safe while the physical world around us becomes interconnected? How can we trust that what we see online is accurate and real? And, is it even possible to learn effectively in a virtual realm when our schools and universities have been closed for months?
Each of these questions has complex dimensions and trade offs, but that doesn’t mean people around the world aren’t trying to tackle them. In today’s View, we take a look at a few examples of solutions being advanced that harness new tech, while also addressing these kinds of concerns. Read on for a deeper dive about a contact tracing app in Germany, an algorithm that sniffs out deep-fakes, and the state of on-line education…
SOLUTION 1: Germany paves the way for privacy-focused contact tracing apps in Europe
Since the start of the Covid-19 outbreak, lawmakers around the world have proposed different ways in which technology can be leveraged to quell the spread of the virus. On one front, technology experts and epidemiologists are scrambling at full force to find a vaccine to serve as the magic cure, while on the other side the focus is on prevention through electronic contact tracing.
Although electronic contact tracing had numerous critics at the start of the pandemic due to privacy concerns, European countries such as Italy, France, and Germany have moved ahead deploying their own contact tracing phone applications. One outstanding success has been the German’s “Corona Warn App, which was downloaded 6.5 million times on its release day.
The German Corona Warn App has become a central talking point over the past week as it seeks to find a balance between the privacy-savvy german people, GDPR regulations, and the effective tracing of positive cases. The good news is that even through the lens of meticulous security analysis, the German application has been praised for being the right approach to privacy-minded contact tracing.
First and foremost is the fact that the app is fully open-sourced, meaning that anyone can see, review, and even contribute application code. If there were any malicious data collection tools or code vulnerable to hackers, they could be seen and fixed by anyone – experts, contributors, or hobbyists alike. Another great privacy focused feature is that the application relies on bluetooth technology to track cases rather than the phone’s geolocation tracker. Since bluetooth is designed to work at short distances of 2 to 3 meters, and the application doesn’t use any geolocation features, it is unable to track or use your location in a map.
So how does the app know you have been in the proximity of someone who tested positive? Through bluetooth, the app exchanges reference numbers with the people you have been near to for 10 to 15 minutes – the nature of the reference codes does not reveal information from your phone either. If your bluetooth detects you have been near someone for too long, it then proceeds to store the other person’s reference number in your phone’s storage for two weeks. By storing the data in your phone itself, rather than in a centralized database, the app further protects the user’s privacy.
When someone tests positive, the person can then decide to upload this information to the app letting the application know about an affirmative test case. Subsequently, the application sends the reference number of the infected person to all phones, who can then cross check the infected person’s numbers with the ones they have stored within their phone. If you have the infected person’s number, the app will let you know that you have been in contact with someone with coronavirus, without revealing who that person actually is. Because of this, the app unfortunately can’t tell you in real time if you are next to someone with the virus, but by doing so it does ensure the privacy of the people using it.
Germany has also made the application fully voluntary, and neither organizations or citizens are obliged to download it. Furthermore, even if you test positive, you are not required to upload this information to the application.
We’re aware of the positive design choices of the Corona Warn App, but it is still good to be aware of possible ways information could get leaked. One possibility could be through intrusive third party applications. An example of this, would be the notorious data collecting application Tik Tok, which has been reported to lurk around your phone scraping available data – even outside the app. This could be an issue if the Corona Warn App stores information that third party apps could deem interesting.
Another issue emerges when a user wants to inform the app it has tested positive. Usually this is done with a private barcode given by a medical institution that confirms you have tested positive. The problem is that not all medical institutions issue this bar code at the moment, so to confirm a positive case, some users would have to call a central party run by the German Telekom to confirm a positive case. This in itself might reveal the identity of the person through their phone number or be uncomfortable for some users
All things considered, the Corona Warn App is a great response to control and speed up the recovery in Germany and Europe and appears to be, for the most part, secure and private. Research from Cambridge has indicated that contact tracing apps would be very effective if 60 percent of people downloaded them. Right now, Germany is between an 8 and 10 percent download rate.
SOLUTION 2: Deep-fake videos have no chance against this detection algorithm
It’s clear that technology can improve life with simple innovations. With that in mind, sometimes technological innovations can act like double edged swords. A clear example of this that has been in the news for the last couple of years is known as the infamous ‘deepfake’ trend.
This trend started to gain a negative reputation a year and half ago when there was quite a stir in the news around a video of ex-president Barack Obama calling the current U.S. President a ‘dipshit’. This was only one in a collection of fake Obama videos where researchers trying to prove a point, or random internet users with more malicious intent, altered and replaced Obama’s authentic facial expressions and uttered words, while keeping the sound of his voice and look of his facial features intact. The result was quite perturbing as people reacted at the uncanny power to create fake videos of the ex-president appearing to say things he never said.
This process of using AI technology to replace an existing video with someone else’s words or actions is known as “deepfake” videos – a relatively similar process to that of fun and friendly Snapchat ‘face-swap’ video filters. Even though this field has had roots in universities and organizations since the start of the century, the potential for negative abuse has brought numerous concerns. To the human eye, most deepfakes are usually easy to spot, but there still has been a constant growth of quantity and quality of these videos. The great news is that detectors are as good, or even better than the technology used to create the videos. Over the last year, researchers have been competing to release increasingly powerful detection algorithms.
One such tool named FaceForensics++ is ultimately changing the field of deepfake detection by publishing incredible results with an astounding rate of detection. Indeed, it was able to detect 99 percent of raw uncompressed video, while also being able to detect between 80 and 95 of high quality compressed videos – the ones usually found in social media sites like Youtube or Facebook – depending on the technique used to create the deepfake video.
In a separate experiment, the same researchers asked 200 participants to classify a set of videos as being fake or genuine. In some instances, the participants fared quite well, but in others some participants only achieved a 50 percent success rate – basically as good as randomly flipping a coin. It’s evident that these detection tools will be pivotal to retain the integrity of news, especially during election campaigns.
Already under heavy legislative scrutiny, the social media site Facebook seems well aware that even more controversy could arrive its way if it’s not able to control these fake videos. For this end, Facebook has just released a huge database of 100,000 deepfake videos for research purposes. The aim is to improve the future detection algorithms to spot any deepfake videos online.
Although Facebook has rightfully received criticism for some of its practices, this database will become indispensable for the future of automatic detection tools that could be deployed by all social media sites for the pursuit of veracity.
SOLUTION 3: Online Education Today
Even though researchers are working hard to increase the trust we have in the online content we see in our day to day, there is no question about the impact that this technology has had in our learning and discovery. With a simple internet connection we now have access to millions of different information sources. It has made finding knowledge easy for anyone by facilitating online education and access to resources.
Online education has been around for more than a decade providing students from all ages courses that provide extra knowledge in a field, or even a brand new skill. With uncertainty over reinstating normal education, e-learning will have to keep innovating to be a more viable source of knowledge for all age groups. So far, online education platforms like Coursera, Udemy, and the free and non-profit Khan Academy have reached an all time high in user base and course subscriptions. But are these platforms really changing the way we learn – or more importantly – will they replace our old education systems?
One of the biggest advantages to online learning is that it can really help students achieve mastery of a subject. This is because online learning can be tailored to be an individual experience – a student can decide to move forwards once he or she feels that they have understood the subject rather than being dragged along with a whole classroom when they don’t feel ready. Studying for mastery seems a more reasonable approach to learning: Khan Academy’s impact page features full studies that show higher grades and understanding across subjects in schools that implemented online learning assistance through their website. Online education also has the advantage of not being bound by time or space. Anyone with an interest in a subject can learn anywhere and at any time.
For teachers, technology has also been revolutionary by allowing the creation of better visual and explanatory content. Online educators like Grant Sanderson (also known 3Blue1Brown) have created mesmerizing graphics that allow viewers to intuitively understand some of the most complex mathematical concepts.
Although technology can have beneficial effects at an individual level and has given teachers more tools to work with, online advocates like Sal Khan still admit that it lacks something: physical interaction. “The learning experience is fundamentally a human connection experience,” Sal Khan said, and he is certainly right about it. Even though online education has allowed for more content to be available at any time, there hasn’t been a viable innovation yet that has replaced the physical classroom. Futurists might imagine the VR classrooms of tomorrow, but so far that vision has not come to fruition.
More importantly, online learning would also have to overcome the technology barrier. Students with unreliable internet connection or lack of computer access would struggle to participate in digital learning. Responsible governance can surely help close this gap – an example that can be seen with New York city distributing 300,000 connected computers for students around the largest school district.
Human-centered technology: the way forward
One general theme that seems to be the main driver of these innovations is the human-technology interaction, or the way we design technology centered on human beings and the human experience. All in all, human-centered innovation with well-being, privacy and veracity in mind can serve us effectively, especially during times of crisis, if we use it responsibly.
About the Author
Pietro Vigilanza is a staff writer at the Optimist Daily, covering Technology, Science and AI. He is based in Rotterdam, Netherlands.