Let’s face it, ‘miniscule bureaucratic innovation’ doesn’t really excite most of us who work in higher ed. But if you’re reading this blog, then Algorithms of Education: how Datafication and Artificial Intelligence Shape Policy may be the new book for you.
When I hear about AI reshaping education, it’s usually something along the lines of robot assistants in the classroom. The reality is that possibly the biggest shift in education is not happening at that intense interface between teachers and students. It’s happening through algorithmic decision-making at the governance level.
We already know that the stakes have changed dramatically since the power surge in modern AIs and their overlap with massive datafication. But are we prepared to accept the fact that synthetic decision making will highlight and amplify our human failings? On the flip side are we willing to accept that synthetic decision making will offer us options that we could never come up with on our own? That data-driven decisions may contradict political rationalities to which we’re deeply wedded?
I know I’m not ready, but this combined human and machine governance seems to be already upon us.
What is governance?
Well firstly, governance is not government. Or it is not only government. As a layperson, I’m already way out of my depth trying to explain it (cue Governance experts’?) but here goes…
Governance is supposed to be the stuff that makes sure organisations do what they should be doing. It’s a system and structure that helps organisations manage themselves. Someone who works in governance might say it gives managers the tools to run things…
legally, ethically, sustainably, and successfully, for the benefit of stakeholders, including shareholders, staff, clients and customers, and for the good of wider society.
Chartered governance institute of UK and Ireland
Another view is that it is like Hobbes’ giant sea monster Leviathan: complex, impenetrable but all consuming and essential to maintain control of the rabble.
Network governance
The most prominent form of governance we experience today is ‘network governance’. This is where some control has shifted away from being state-centric. This control is then distributed to a diverse range of organisations and actors like consultancy firms, philanthropic groups and tech companies. Think the Paul Ramsay and Gates Foundations.
Data-driven public policy also forms a core part of the rise of network governance. This datafication shapes the outcomes we desire and the methods we use to reach them. It relies heavily on interoperability and standardisation of data production and sharing across networks. Massive datafication is essentially the reason that algorithms have gotten into the governance mix.
What is synthetic governance?
Synthetic governance is speculated to be the next important shift. It is the synthesis of human thinking and values on the one hand and new forms of computational or ‘non-conscious’ cognition on the other. Basically, the nonhumans are getting involved. This idea isn’t new but it’s only coming into really sharp focus in the era of AI.
How does it work?
Synthetic governance is characterised by three factors:
- Human classifications, rationalities, values, and calculative practices.
- New forms of computation, what we might consider to be nonhuman political rationalities, that are changing how we think about thinking.
- The new directions made possible for education governance by algorithms and AI.
These ‘new directions’ are where humans cede our agency. They are drawn out of the kind of AI potential demonstrated by DeepMind’s AlphaGo and GATO.
The creative potential of AI
This is where it gets rather spooky but also extremely interesting.
More than computation: This is when we realise that deep learning is not just a system for doing really complex calculations or guiding decisions towards optimised outcomes. This is really different to earlier forms of AI like IBM’s Deep Blue where optimising a rule-based approach was the goal.
Meta-learning: Deep learning actually learns from it’s own ‘hidden layer’ network of algorithms, what some have called “meta-learning”.
Learning to learn: This is where deep learning engenders a move beyond learning from experience towards learning how to learn by context. This becomes its own way of knowing by “reasoning through and with uncertainty“. It is in this uncertainty, that the creative potential of AI starts to surface. In uncertainty, machines may find new ways of doing things as opposed to the ‘optimal’ way doing things.
What does it mean for higher education?
Data science doesn’t yet hold a central place in education policy and governance but clear changes are already occurring. This could be simply because datafication is at the core of modern higher ed:
acquiring information about the performance of students across a range of fields and then issuing with them credentials underwriting the authenticity of that information.
pg.3 Algorithms of Education
Facial recognition is one area where we already see the biggest growth in synthetic governance. For this, we are actually back in the classroom at ‘that intense interface between teachers and students’. Apart from transforming students’ faces in to statistical data, these systems also have a huge impact by automating and conflating administrative roles with learning. This is one example where we can see one potentially sinister aspect of synthetic governance in education creating new norms for human behaviour.
For some of us that example could be seen as something that transgresses societal norms. What it really shows is how contemporary AI is creating new patterns, new markers of normality and aberrance besides which we could be calibrating ourselves. For such a situation, we may want to avoid being tied to the old dichotomies we have about AI. Instead, there could be space for more critical awareness of the new knowledge, new values and new decision making processes that are forming education policy and governance.
Learn more about synthetic governance by viewing the conference keynote Recording available from #DigiEduGov22 | Digital Education.
Gulson, Sellar, S., & Webb, P. T. (2022). Algorithms of education : how datafication and artificial intelligence shape policy. University of Minnesota Press.