The Biggest Risk of AI in Education?

It makes us more efficient at creating ineffective learning experiences

Maybe the biggest lesson we have learned from the failed promise of online education is that, for all of the hype, new technologies usually imitate and automate established ways of thinking and doing.

Online education has failed to deliver on its promise to transform education because the technology that we have built to deliver it reproduces broken systems of teaching and learning.

MOOC platforms, technologies aimed at higher ed, corporate L&D platforms and consumer-facing platforms like Udemy & Teachable: all of them reproduce a broken, content + regurgitation model system of instruction misaligned with what we know about how humans learn.

In order for AI to be truly valuable, those who build AI-driven education tools must understand and make better use of learning science.

Rather than building technology to detect AI-generated essays, or asking AI to help us to generate content & knowledge-check assessments more efficiently, what if we ask instead: what are the optimal conditions for human learning, and how might AI help us create these conditions?

Here are three examples of what this could look like in practice:

---

Example 1 - AI as a Tool for Intrinsic Learner Motivation

Research shows that without intrinsic motivation, learning gain (i.e. acquiring applicable knowledge & skills) will never be optimised.

Research also shows that learners try harder, persist & learn more when they have a sense of belonging and are not distracted by a sense of isolation, alienation or exclusion.

What if we built AI that could:

• Deeply understand the learner and actively address stereotypes which might impact their motivation.

• Generate a curated range of examples, case studies, resources & activities to reflect diversity of thought, experience & perspective on the topic, including gaps in the research caused by cultural, political, racial or other biases.

Example 2 - AI as a Tool for Foundational Knowledge-Building

Without the ability to recall & process accurate foundational information, the learning process is futile.

Research shows that recall increases in both quantity and quality when learners are stepped through carefully designed worked examples which show experts making decisions & solving problems, then have a go themselves. The effect is especially high for novices.

What if we built AI that could:

• Generate carefully structured worked examples through a simple prompt.

• Provide an infinite loop of explanatory feedback until the learner demonstrates mastery.

Example 3 - AI as a Tool for Deep Understanding & Expertise

Without an opportunity to apply it, memorised information is too abstract to be useful. Without conceptual context & explanation, experience is just a collection of memories. 

In combination, information + experience is proven to produce usable knowledge and/or deep understanding.

What if we built AI that could:

• Coach learners thought repeated cycles of so-called "productive agency" where learners are prompted to create something using their knowledge, e.g. a text, video, image, audio or object - share it & get feedback until mastery is achieved.

• Design and deliver Deliberate Practice - short sprints of intense, intentional action followed by discrepancy and explanatory feedback to drive a learner to mastery.

I’m in the process of generating a list of how AI could help us, finally, to scale the positive impact of learning science. What would you build, and why?

Happy designing!
Phil 👋

🎓 References

Atkinson et al. (2000). Learning from Examples: Instructional Principles from the Worked Examples Research. Review of Educational Research, 70(2)

Bransford & Schwartz, It Takes Expertise to Make Expertise: Some Thoughts about Why and How in Ericsson, K. A. (Ed.). (2009). Development of professional expertise: Toward measurement of expert performance and design of optimal learning environments

Cohen et al. (2006), Reducing the racial achievement gap: a social-psychological intervention, Science (September, 2006)

Dougherty, D., The Maker Movement, Innovations: Technology, Governance, Globalization (2012) 7 (3)

---

Want to learn more about the science of learning? Great!

• Sign up to my AI for Educators self-paced, video-based course (coming Feb’ 2023). 🚀 The first 100 sign-ups get free access to the Learning Futures Digest for 12 months!

• Apply for a place on my Learning Science Bootcamp.

• Subscribe to:

  • the Learning Science Digest: a monthly summary of the most important learning science research, translated into learning design practices.

  • the Learning Futures Digest: a monthly newsletter exploring the relationship between between emerging technologies and education.

• Access free guides and other info on my website.