In talking to people about this, I've found that many educators are still not really sure just what AI and predictive algorithms mean for education, how LMS companies do data mining, what the difference is between machine learning and traditional statistical analysis, etc. etc. Over the past five months, I've been collecting online materials on these topics, so in this special "InstructureCon Edition" of my #datamongering round-ups, I've listed what I see as some of the most valuable resources people can use to learn more. Read on:
1. Instructure: Plans to expand beyond Canvas LMS into machine learning and AI by Phil Hill. This blog post is where I first learned about the big shift at Instructure, and you will find extensive quotes from Instructure's new CEO, Dan Goldsmith. This is a must-read for anyone whose school is using Canvas LMS:
2. Despite Dan Goldsmith's claims about Instructure's database, there is nowhere near enough data in Canvas to model real learning by real students. What kind of surveillance will be required to get the actual data required? China has a Next Generation Artificial Intelligence Development Plan (NGAIDP) that is bringing full-scale student surveillance to the classroom; there is detailed reporting here from Xue Yujie: Camera Above the Classroom. If you are going to read just one article on AI in education, this is the one to read.
3. For a student perspective, you can listen to the story of Bryan Short, a student at the University of British Columbia in Canada. There is an interview with Bryan at EdSurge: Inside a Student’s Hunt for His Own Learning Data (podcast with transcript), plus an article from the UBC student newspaper that puts Bryan's story in context: Canvas is tracking your data. What is UBC doing with it? by Zak Vescera.
4. Increasing surveillance of students is an issue of great concern for both higher ed and for K-12. On K-12, see this important piece by Benjamin Herold in EdWeek: Schools Are Deploying Massive Digital Surveillance Systems. The Results Are Alarming.
5. For resisting surveillance, and LMS surveillance in particular, you will find a good discussion here: Ethics and LMS Surveillance which is part of #DHSI19: Balancing Issues of Critical Digital Pedagogy containing contributions from Chris Friend and many others.
6. By mining student work to create new products, Instructure is following the lead of TurnItIn, a company which recently sold for $1.75 billion (not a typo). For an overview, see Automating Mistrust by Ben Williamson.
7. Did you notice that Canvas rebranded itself in June as a platform, not just an LMS? (details at the official Canvas blog). For an idea of just what the platforming of education means, here's a great piece, also from Ben Williamson: The platform university: a new data-driven business model for profiting from HE.
8. Matt Crosslin is more optimistic than I am that there is real value in data analytics, and he also recognizes some real pitfalls too; this blog post provides a great overview: So What Do You Want From Learning Analytics?
9. Anyone going forward with algorithms needs to be aware of the dangers involved, and there are indeed many dangers. This resource from MIT points out some of them: AI Blindspot: A discovery process for spotting unconscious biases and structural inequalities in AI systems.
10. Finally I want to close with a brilliant film from sava saheli singh's project Screening Surveillance. The film is not about education, but it's easy to see just how this model employee could be re-imagined as a model student. Leila Khalilzadeh is the director, with a screenplay by Tim Maughan: Model Employee.
So, keep on reading, people! We cannot afford to be ignorant about AI, because . . . The AI Supply Chain Runs on Ignorance.
And if anybody is at the Instructure Engineering panel at InstructureCon on Thursday (July 11) 4:20PM in the Long Beach Convention Center, GB-B, please let me know if they say anything about a data opt-out. I don't know if my question will make the cut or not... but I have not given up hope yet.