Privacy in Digital Learning

As a learner, your privacy can never be perfectly protected. If your learning experience is in any way interactive you must confront the mortifying ordeal of being known whether you like it or not. In order to give you a credential an institution must know to whom they are giving it, which means you can’t be fully anonymous. Beyond these bare minimums, one starts to make some trades. Your university does not need to know your race to educate you, but they do need some details of it if you apply for scholarships benefiting students of colour, and that information will become data that is stored and potentially utilised. Also, demographic surveys given by institutions that have detailed questions can be used to identify students. If an anonymous form says a student is in the music department and identifies as genderqueer anyone who can see the form can already narrow them down to about five. “Choose not to answer” is often an option on these surveys, but as a member of a minority group, picking that option means decreasing your representation in the system, so there is pressure to “show yourself”. There are methods of data collection and storage that can do better or worse to obscure that data, but on a fundamental level it requires students to trust that no abuse of authority will occur, that the current policies around privacy are written and enforced well enough to protect them and will not be changed later in a way that retroactively impacts them. Privacy is the line we draw around what shapes our identity. Our identity is, unsurprisingly, quite personal to us. Crossing that line, therefore, is dehumanising and fundamentally disrespectful of our dignity, which is why an invasion of privacy is wrong, regardless of any material help or harm, and why a principle of minimum required disclosure of personal information should be upheld.

Privacy, ethics, and Educational Needs

Effective education often requires teachers to know things about their students. All student information can be used by a teacher to personalise and optimise learning if they are sufficiently clever, creative, and empathetic. If you know a young student is from a vegetarian family, you know a word problem about the math of buying hamburgers and hot dogs might not be engaging for them. If you know a student is in a house with few books you may offer additional reading support. If you can constantly track a student through a digital learning platform and know how much time a student is spending doing work at a computer in different places, you could see, for example, that they achieve strong results at home, strong results at their friends house, and poor results in the classroom. If you know the student personally you might infer that they are getting other people to do the work for them or that the classroom simply isn’t a conducive learning environment for them and adjust accordingly. Theoretically, every piece of information can be valuable in some way to cater to a student’s individual needs. Privacy, however, is still important. Even if a teacher might benefit from knowing everything about their students, that does not mean that they should or that it is right to probe. Glenn Greenwald’s 2014 TED talk explores the importance of privacy in wider society, in particular dismantling the argument that those with nothing to hide have nothing to worry about. As part of his talk, he highlights how people who feel watched exhibit more conformist, less combative behaviours. Knowing this effect, do we want our students learning under constant supervision? In addition to the previously stated reasons to protect student privacy, without care, an educational system that does not respect it can become a system that forces conformance.

A teacher knowing too much about their students is, in the grand scheme of things, a small issue. Beyond their duty to report, teachers will usually be the only ones to know information that is shared with them. If they start spreading it everywhere they are quite likely to get fired, and, until we get the brain chips going, no one can hack a human teacher, so the data relationship remains largely personal. Giving this data to algorithms that underpin digitally assisted learning platforms offers a new set of challenges. The first is the issue of explicable conclusions. A teacher can make inferences from information, respond to it, and then explain to us exactly how they came to their inferences and what motivates their response. A machine-learning algorithm, however, often operates as a black box, and it cannot offer an explicit report on what information it is specifically responding to or why it does what it does. The design of these models, often by private companies, is many steps removed from the democratic process, leading us to question to whom we are giving such power over the development of our children. The black box of processing makes evaluating the effectiveness of algorithm-assisted learning on an individual level much more difficult. Even if they increase a desirable metric on average, that does not mean they are able to support every student; a 4% increase in average test scores can easily be counted as a success regardless of statistical outliers whose scores dropped by 10%.

As stated, student data accessed by an AI tutor is not contained to an individual. In order to be used it must be processed and stored at a server, and access to that server is politicised. Who gets to see it? School administrators? State or provincial officials? The upper management at the private company running them? What do they get to do with this information? Even if an arrangement that is respectful and effective can be found, if they try to change the rules later, will we have the pressure to stand up to them? Below is a video covering a data breach that leaked large amounts of student information in the Minnesota Department of Education, showing how fragile our privacy really is when so much of our data is stored outside of our hands. It is a general truth of the human condition, in my opinion, that, in all fields, people will take what you give them and use it. Give a tinkerer a gear, they’ll make it spin. Give a parent an iPad, they’ll calm their children down. Give a police officer a gun, they’ll point it at someone. If you give someone information, and incentive sets align with abuse of that information, infringement of privacy will occur. Because of this and regardless of any improvement of efficiency, data surveillance in online tutoring must be checked. This may be seen as stalling progress, but if ethical practice maximised progress and efficiency, we wouldn’t need to talk about ethics.

Universal Design for Learning in Digital Pedagogy

Digital Universal Design for Learning (UDL) is, on a certain level, impossible. Access to technology and digital literacy are themselves not universal, and the barrier set by this membrane of interaction can be alleviated but never fully bypassed, continuing to make digital learning less accessible, serving as another of the many reasons public libraries, which can provide computer access, are necessary to bridge inequality. In other ways, digital learning is particularly well-suited to UDL. UDL is a set of guidelines for developing curriculum and learning activities that cater to the wide variability of learners in a classroom, focused on integrating accessibility accommodations into mainstream learning design rather than separating them into alternate paths. This tunes into an essential and often overlooked aspect of accessibility accommodation: improvements to the lives of the differently abled usually make things better for everyone. Flexibility of deadlines designed for people with chronic illnesses also helps able-bodied people who are in tough times, closed captions for the hearing impaired also help hearing people follow along more easily, and everyone has an easier time moving through a hallway that’s wide and clear enough for wheelchair users. Accessibility is ease of interaction, and whenever it is successful, everyone benefits. The video below gives more specific details of what UDL is and expands on the ways that everyone benefits from universal design.

As a framework based heavily on multimodality of representation and interaction, UDL lends itself well to a digital classroom that can be constructed freely from many physical constraints. Online learning can be delivered as audio, video, text, drawing, or interactive game delivered live, asynchronous, or some combination thereof. The key is that, though online learning can be constructed in many ways, it must be constructed.

Making a video is not easy. Scripting, vocal presentation, filming, recording, graphic design, and editing are all complicated processes that require a certain level of expertise, sometimes in the use of professional tools that are not easily picked up in an afternoon. Corners can be cut, but digital learners are likely also digital livers who are accustomed to high production value content, and a video being unprofessional in its execution can be distracting or harm your credibility. Video and many other digital media have largely moved out of the domain of hobbyists and into the domain of specialists, and because of this, a lot of digital learning is based on finding learning materials that are already available publically. Being able to find publicly available materials that meet the specific needs of your students is predicated on such materials existing, which makes multimodal representation of niche or specific topics considerably more demanding, especially compared to the simple efficiency of a textbook chapter.

All this being considered, a poorly made video that is not perfectly tailored to your lesson still offers learning benefits that would not exist without it; imperfect is not worthless. UDL is not one answer for everyone; learners are not static so neither can be learning; adaptation is at the centre of the framework. In that sense, digital UDL is hardly a new application of the form, but simply a continuation of that adaptation into a new set of media. For all the challenges it provides, we continue learning, we continue building skills, and we get closer to making learning truly universal.