Digital Education and the Future of Learning
Today we talk about digital education and the future of learning. My guest is Ben Williamson, a Chancellor’s Fellow in the Centre for Research in Digital Education at the University of Edinburgh. He wrote the book Big Data in Education: The digital future of learning, policy and practice (Sage, 2017), and is an editor of the journal Learning, Media and Technology.
In our conversation, Ben talks about the many ways data is being extracted inside schools and education systems and reflects on what that might mean for policy and practice. He warns that there are biases built into data.
Citation: Williamson, Ben, interview with Will Brehm, FreshEd, 191, podcast audio, March 16, 2020. https://freshedpodcast.com/benwilliamson/
Will Brehm 1:33
Ben Williamson, welcome to FreshEd.
Ben Williamson 1:35
Thank you very much, Will. Thank you for inviting me on.
Will Brehm 1:38
Can you give me a sense of the reach today of education technology?
Ben Williamson 1:44
Okay, so I think education technology has really expanded across education systems in recent years. So, we now see a tech in the early years, we see it right through primary and secondary schooling, through to university and even on into lifelong learning. So, there’s been a kind of a growth of what some people call a global education industry, that increasingly serves pretty much every function of education. So, we have things like attendance monitoring in schools and universities through a smartphone, for example. We have the fairly conventional technology-enhanced learning and teaching resources. And more recently, we’ve seen the development of more data-driven type technologies, including learning analytics, and adaptive learning software, and even perhaps at the most extreme end, Artificial Intelligence (AI)-based learning and teaching assistance. So, some people are starting to talk about we might have AI-powered smart schools or intelligent campuses and so on. But when I think about education technology, I think, on the one hand, we have these kind of spectacular examples of things that are going to allegedly transform the classroom. But most education technology, I would say, is much more mundane. It’s kind of sunk into the background. It’s things like learning management systems, or student information systems, or even large-scale data infrastructures for gathering national level student data sets. For example, we have a new national student data infrastructure for higher education currently in development in the UK that’s due to roll out later this summer.
Will Brehm 3:40
What kind of data is that system trying to analyze or to gather?
Ben Williamson 3:46
In the UK, in higher education, we’ve always collected extensive amounts of student information from their background demographic details, to all their grades and so on. The idea now is that the upgraded infrastructure for doing that will allow data about students to be collected throughout the academic year. That data will be centrally held and made available to various different organizations that need it. It also comes with extremely enhanced data analytic capacity. So, each university in the UK will have access to a new data platform for doing data analysis on their own students, for doing competitor analysis, and benchmarking, and comparison, and so on. So, we’ve seen this kind of intensification of what is possible to do with data in higher education, which is being made possible by this infrastructure upgrade that’s ongoing at the moment. And like I say, this is a fairly invisible, kind of hidden part of the architecture of education. It’s not the kind of thing that students see or that teachers see in the classroom. It’s not AI robot assistance or anything but it has a potentially really significant impact on the way in which education is understood. And the way in which students progress is measured. The way in which performance of courses, faculties, staff themselves is also assessed.
Will Brehm 5:16
It’s like everything can now be measured. And everything is, the government’s, or school systems, or EdTech companies are trying to figure out new ways to measure pretty much every part of the educational experience. And I mean, it really has changed quite a lot. I’m thinking about the EdTech when I was a student in grade school and it was things like putting computers into classroom. But that’s a long way off from learning analytics to try and measure how I best learn as a student. It’s a very different dynamic here.
Ben Williamson 5:51
Yeah. So, we’ve moved from ideas about data collection largely being about assessment, which is very kind of temporarily bound. It happens at long intervals. And now we have ideas about analytics, being able to conduct a kind of real time assessment without the necessity of testing. So, there’s a kind of constant assessment and evaluation, or kind of a progression monitoring of the student that’s available through the application of various different kind of analytics platforms. So, I don’t think we see any great break with the past. We still obviously have assessments and so on, but what we see is perhaps an intensification of assessment, and also transformations in the way that assessment can happen. And perhaps most significantly, the capacity for these analytics to themselves learn from the process of assessing students. So, they learn about the students’ progress, make predictions about the students’ future progress based on comparison with a vast database of other students, and then have the capacity for making feedback, or producing recommendations for students themselves or teachers on how better to proceed. So, yeah, there’s a kind of continuity from the past in terms of continuing to assess students but also an intensification and to some degree, a kind of transformation of what is it now in the assessment gaze?
Will Brehm 7:31
And what is the logic behind the intensification? Why do we need more digital data in our assessment of students and in the educational management systems? What’s the logic driving that need? What’s the reason for it even existing in the first place?
Ben Williamson 7:50
I think this is largely being driven in the UK by kind of “performance logic”, the logic that if we want to see improvement at the level of schools or at the level of universities, then the best way of doing that is by actually tracking their current performance and comparing that with historical performance datasets, and comparing it with other kinds of institutions, and then using the insights from those analyses to feed back in to what the institutions themselves are doing. So, although the technologies themselves are often coming from a kind of global industry, the drivers for that are often much more kind of typically governmental led. This is about governmental logics of performance management of institutions and the logic that data can be a source of constant improvement and organizational self-learning and so on.
Will Brehm 8:49
And has it actually happened? What do we know about the success, or impact, of all of this digital learning on student learning?
Ben Williamson 8:59
I think the question of whether there’s been a notable impact or success on student learning is a really open one. I’ve yet to see compelling evidence that the gathering of all of this data leads to increased or improved student learning. I suppose there might be other ways of thinking about the impacts. What we might say is that there’s an impact on the way in which pedagogy is conceived. And so, the drive to make more use of analytics and adaptive platforms goes along with ideas about personalized learning, and individualization, and customization around the student and so on. So, the impact there is about introducing a degree of automation into pedagogic routines. I don’t think we’re talking about entirely replacing teachers but what I do think is that we’re on a trajectory towards the kind of robotic augmentation of some aspects of the kind of pedagogic routine of schools. And one potential effect of that could be a kind of reductive account of what learning is. If learning is only what can be seen and observed and recorded by a particular kind of analytics technologies then that seems to shut out other kinds of forms of teaching, or other ideas about the value of different approaches to teaching, and so on. So, I think maybe we can see the beginning of some impacts on the way in which teaching is performed. And the way in which we conceive of what kinds of pedagogies are valuable. And the danger is that we only conceive of valuable pedagogy as being those that can be measured and counted and quantified in those kinds of ways.
Will Brehm 11:10
Right. It would miss all sorts of other elements that are more qualitative, more difficult to define. The other thing that I find so fascinating is that with data, and big data, there’s this sort of common assumption that being able to measure something and have a data set is in itself objective. And sometimes it seems like it’s a bit more subjective by selecting which proxies you want, or which measurements you want, or which measurements you value. So, in EdTech, and in the sort of digital technologies in education that you’ve been looking at, are these data points objective? Or is there more to it?
Ben Williamson 11:51
I think there’s a really common perception in the education technology sector or in the technology sector more widely, that data do provide a kind of objective, realistic, impartial, and neutral representation of what’s really going on in classrooms. And that the data are really accurate proxies of particular kind of cognitive processes that are involved in learning and so on. I think what that kind of realist perspective overlooks is that the systems for collecting data always have to be built. They have to be designed, they have to be engineered, they have to be tuned, they have to be checked, somebody has to do a load of statistical waiting. And so, all of those very, very human and social activities actually shape the kind of data that you eventually get. Which is not to say that you might not get useful data. But I think we need to recognize that the data are not simply these kind of objective “God’s eye views” of what’s really happening in classrooms. And the data is also extremely partial. That the software can only measure what it’s been told, or programmed to measure. And there may be all sorts of other things going on that are contributing to a particular learning experience, or to a particular moment of progression for a student which are completely outside of the view of the software that’s doing the recording. So, I think we need to be really cautious about these ideas of data and objectivity.
Will Brehm 13:26
So, you said earlier that a lot of this sort of comes from government mentalities, and performance management, and assessment, and sort of that logic entering schools and education system. But are there other actors involved in this EdTech world producing all of this data beyond governments?
Ben Williamson 13:48
For me, what’s really intriguing about our current moment, is that we seem to have governments and education technology companies speaking often the same language, and pursuing pretty similar goals. So, lots of people working in the kind of education policy space, talk about network governance, and policy networks, and so on where policy and governance is no longer just the preserve of government departments and ministers and policymakers and so on, but actually you have these really complex mixes of government centers and education businesses and think tanks and consultancies and so on. And I think that’s a really useful way of thinking about what’s happening in our particular kind of education technology moment. That there are multiple different organizations and actors from very different sectors coming together around shared aims and visions and aspirations for what a kind of technology-enhanced education could or should be. So, some of the other big players for example, I’d include Pearson, the world’s biggest education company, which recently announced it was ditching its traditional textbook business and moving to a digital-first platform business model. So, it’s going to be delivering all of its teaching resources on a kind of on demand subscription streaming model. So, it’s talked about becoming the Netflix for education. And Pearson is also pushing its Artificial Intelligence learning and teaching assistants, which it calls AIDA. So, Pearson, I think, is a big influence on education technology at the moment, and particularly this move to creating products that are fueled by student data and are responsive to students on the basis of learning from what the students do. Some of the other organization, I mean, Google clearly is a huge actor, G Suite, and Google classroom are now used by millions of students, and are incredibly influential arguably in introducing young people to the kind of Google world of services and so on. And then you’ve got organizations like the Gates Foundation, of course, which has been pushing education technology and personalized learning for a long time. And then newer organizations like Mark Zuckerberg’s Chan Zuckerberg Initiative, what some people call a for-profit philanthropy because it mixes venture capital investment with more traditional models of grant giving, and so on. I mean, the Chan Zuckerberg Initiative, despite the fact that it’s only 5 years old, has already given hundreds of millions of dollars to support its own vision of personalized education and what it describes as the use of “learning science” and “learning engineering”, as ways of improving education and making it much more scientifically rigorous and evidence-based. Then of course, there are multiple other organizations but I think that gives some sense of it, we’ve got lots and lots of government interests, and we’ve got the think tanks and the consultancies, and then we’ve got these really big, increasingly global organizations all largely singing from the same kind of hymn sheet.
Will Brehm 17:26
So, beyond the idea of all of these for-profit companies infiltrating all aspects of the education process, from early childhood all the way through lifelong learning, as you said, you know, beyond that sort of profit motive and privatizing various parts of public education, why else should we be worried? Some people might say, having all of this technology, having access to G Suite, having the ability to use computers and gather all of this data is valuable. So, what would be some of the other worries that you have when you look at this plethora of actors in this space?
Ben Williamson 18:05
I suppose one of the obvious worry is that there’s pretty compelling evidence now that students’ personal data is leaking out all over the place. So, it’s a really important project in the States on school cybersecurity, which is mapping out now hundreds of data breaches and ransomware attacks, and so on. So, I think that’s a clear concern. I think one of the others is that we’re often seeing for-profit organizations now increasingly influential in shaping the direction of public education itself. So, we actually have private sector business models, which are concerned with developing new data-driven technologies, and then finding market niches for those products, eventually, and over time, reshaping the way in which schools are organized and the way in which universities are run and so on. So, I think we need to just keep our eye very, very closely on those kind of wider and significant shifts in how we think about the nature and the value of education itself, which I think is being changed and shaped by these other kinds of organizations, which have become almost like kind of shadow authorities in public education itself. They are engineering products, but they’re also re-engineering education itself in some sense. If we think that data invites students to look on themselves in new kinds of ways, or it invites or encourages teachers to look at students and understand students primarily through their data traces, then I think we really do need to consider the way in which this is shaping different kinds of subjectivities in classrooms. Different ways of relating to oneself or relating to others. Teachers are being encouraged to look into the classroom through a kind of data gaze, as David Behr calls it. To see students represented first and foremost through their numbers, because it is those numbers that,as it were, count the most and on which teachers themselves will be held accountable. They’re responsible for student performance in multiple ways and the data seems to indicate whether teachers are doing that. So, I think these are all incredibly significant issues in terms of a kind of re-engineering of those people who inhabit classrooms.
Will Brehm 20:45
There’s this idea that you point to called “precision learning”. What on earth is precision learning?
Ben Williamson 20:54
So, I see precision learning or precision education as the kind of extreme end of education technology at the moment. It’s where EdTech software engineering meets the sciences of learning and cognitive science and neuroscience, and even to a certain extent biomedicine. So, I think it’s come from a number of different kind of sources. Firstly, we have organizations like the OECD, an extremely influential global organization pushing for greater use of scientific insights into learning. And it claims that policy-relevant scientific insights are coming from cognitive science, from psychology, from neuroscience. And in particular, it argues that these insights are being generated through sophisticated lab technologies, and AI, and machine learning that’s all being used to analyze really large scale quantities of student data. Then we have, in addition to those kinds of bigger drivers, we have new spin-out organizations, which are developing things like wearable brain scanning headsets for monitoring the kind of activation and oscillations of the brain during learning experiences. So, EEG headsets, and head nets, and so on. And there are a number of startup organizations that have developed those kinds of technologies, which are now on the market. They’re available and they’re being used in various research projects to try and explore the kind of brain-based or the kind of neural substrates of learning as it were. And then precision education is sort of completed, if you like, by learning analytics, or developments in learning analytics with the claims that it’s possible through the gathering of these enormous data sets about the brain, or about cognition and so on to identify patterns in these huge data sets. And on that basis, to then make precise, personalized interventions and recommendations. So, it’s taking the logic of personalized learning and saying, well hang on, let’s make personalized learning brain-optimized, or cognition-optimized based on the masses of data that we have about both individuals, and huge populations against which they can be compared. So, there is one company based in Boston- came out of research in neuroscience at Harvard University- that I saw recently had announced what it called a “brain optimized personalized learning platform”. So, I think that’s the kind of direction we’re heading in in terms of a kind of precision education model.
Will Brehm 23:55
It’s like a bad Stanley Kubrick film of the future dystopic world of education it seems like in my mind. You know, these students wearing these wearable technologies so the data is gathered and analyzed instantaneously to say exactly when a student is or is not learning and to what effect. It seems like science fiction. It seems like a pipe dream, in many ways, like it would never exist. But yet we have all of these companies that already do exist that are doing it and big universities, big famous universities are putting a lot of money into it. So, it’s not just a joke, as I sometimes think about it.
Ben Williamson 24:38
Yeah. The Chan Zuckerberg initiative -to return to them- recently announced that they’d given I think it was $2.9 million to a particular lab at the University of California, San Francisco, which is developing these very technologies and actively describes itself as the Precision Learning Center. These kinds of labs are where, in part, the future of education is getting invented, you know. It’s trying these kinds of technologies out, seeing what kinds of insights we can get from them, experimenting with adaptive algorithms and machine learning to see then, whether the sensor kits and the analytics stitch together can ultimately provide really precise, personalized, adaptive types of recommendations and interventions.
Will Brehm 25:39
It’s almost like we’re now not even talking about teachers anymore in the very process of schooling and learning.
Ben Williamson 25:45
Yeah, I think, well, maybe we’re talking about a very different sort of teacher. I mean, there are certainly some out there who talk of genuinely trying to automate the task of teaching, I think, the more realistic view is that we will see augmented-type teachers, some of whose professional responsibilities are delegated to artificially intelligent machines. I think that raises all sorts of questions about teacher professionalism, and teachers capacity to challenge the kind of recommendations and insights of these kinds of automated assistants in the classroom, and it will have to do an awful lot of work to help prepare teachers for their automated teaching assistants entering the classroom. Yeah. I don’t think it will have teaching without teachers, but I do think we’ll have teaching with human and non-human teachers in the next 5-10 years or so.
Will Brehm 26:49
Gosh! Could you imagine being a teacher and having this precision learning data coming down at you saying, this is how to do it. How do you fight back? How do you say, Actually, I have a different opinion than your opinion based on a million data points?
Ben Williamson 27:04
I mean, I don’t know how to fight back, really. I’m aware that there are some activist groups, particularly in the States, really challenging some of these developments, you know, teachers and parents. I think there have been some interesting, specific examples of students walking out of schools that have really invested heavily in personalized learning software. So, it suggests that there’s a kind of an awareness amongst certain groups of educators and students themselves of the real limitations of these kinds of technologies despite all of the hype, and the optimistic, utopian promises made about it.
Will Brehm 27:51
Yeah, exactly. Now, is there, you know, another thing, at least in the States, that’s becoming quite popular and common, I would say -I mean, I even did it- is DNA sequencing for a small fee by companies like 23andme and I’m pretty sure there’s other ones. And, you know, I did it, and then only later did I really start thinking, oh, my gosh, there’s all this data now that is owned by, I think, the former wife of a Google executive or founder. So, it’s some private company that owns all this data about myself, and who knows how it will be used, and to what effect. But, is DNA sequencing and the data that comes out of that being integrated into some of these other EdTech digitalized sort of platforms that you’re seeing emerge at this moment.
Ben Williamson 28:43
Thankfully, we don’t see DNA-based education platforms yet but this is an emerging topic. And it’s, I think, an extremely contentious and controversial one. So, 23andme is already involved in education research. And so 23andme is a collaborator with behavioral geneticists and what are called geno-economists, who are doing very large scale studies on the associations between DNA and educational outcomes. So, a study that was published a little under two years ago had a sample of a million people who had voluntarily donated their DNA, either to 23andme in the States or to the biobank in the UK. So, we now have these extremely large genetic studies going on, looking for complex associations in patterns of DNA and eventually educational outcomes such as attainment. When you spat in a vial for 23andme you would have filled in a background questionnaire and educational attainment is commonly collected in those kinds of questionnaires. So, that makes it possible to do these kinds of studies. To say, well, okay, we’ve got millions of samples and we can now identify as the scientists on this particular study have identified 1,200 particular kinds of genetic markers that are associated with how long you spend in school or university. So, I’m not aware of these kinds of findings yet being taken up in either policy, or in education technology development but there is a very live debate at the moment about whether it might be possible to use genetic data to assess students strengths and weaknesses, and then to make personalized learning decisions and predictions based on that. It’s important to acknowledge that not all the people involved in this kind of science agree that that’s a desirable outcome but others do. They think if we can analyze DNA and identify that some students will succeed in school and others won’t, then we have a responsibility to vary what they receive in school, rather than pursuing a one-size-fits-all curriculum where someone inevitably will fail. The view that you can use DNA to do personalized education is certainly not shared by all but it’s, I think, a really live topic which we need to focus on over the next coming years. And that’s what I’m hoping to focus my future research on.
Will Brehm 31:49
I mean, it’s a slippery slope to eugenics, it seems. I’m sitting here at UCL, which is where eugenics was created. And there’s a live debate happening right now on campus about how do we remember certain scientists who really advanced and promoted the use of eugenics in policy and practice. And people often think of it as a historical debate but it seems as if there actually is some scary, contemporary parallels happening as well.
Ben Williamson 32:23
Yeah. I think it’s really important that we’re paying attention to really important and dangerous ideas about eugenics again, at the moment. I mean, obviously, just in the last couple of weeks, we’ve had major controversy in the UK about eugenics thinking inside of central government. At the same time, I think we need to be cautious before hitting the eugenics button in relation to some of the studies that are going on. It’s quite clear that projects, which claim that you can personalize education based on DNA could lead to horrible forms of discrimination, which we absolutely must counter. But the majority of the genetic scientists working on this kind of research, really are not interested in trying to intervene in that kind of way. Now, I think we should still look critically at the kind of findings they’re producing. But I think it might also be unhelpful to approach all of this kind of research in the first instance, as if it is all you know, yet another step closer to eugenics. So, certainly, in the research I’m hoping to do in coming months, perhaps years, is to try and trace much more closely, what are the actual processes through which this new genetic knowledge is being produced? How is that knowledge affected or influenced by the kind of infrastructure of technologies that is doing the data collection and analysis? And then what kind of work goes on to translate those findings into a kind of policy-relevant language, if at all? Because, you know, these findings might not be taken up by policy at all. And certainly, some of the scientists are actively resistant to the idea that there are policy implications of their work. But at the same time, we need to recognize that there are certain high profile commentators who get invited to write high profile columns and so on who are already saying, we should make use of genetic data for all sorts of reasons. And some of those people sit pretty firmly in the right wing and have highly conservative views. So, this is a really contested space, but I think we need to be really careful about teasing out all the various different perspectives and arguments that are being made.
Will Brehm 35:05
Ben Williamson, thank you so much for joining FreshEd. Really a pleasure of talking today and when you get some more research done on genetics, and education, and precision learning, please come back on and share your findings with us.
Ben Williamson 35:19
Thank you very much, Will. It’s been a real pleasure being on the show.