Motivation Blindspot

Hey 👋

Hope your week’s going grand. Today, another AI-focussed snack for ya…

Big idea 🍉

The idea of a personalised learning future—where every student has their own AI tutor, and outcomes soar—is experiencing a resurgence. However, despite the hype, this vision is likely to face significant hurdles, potentially floundering before it truly takes off.

This is not because LLMs won’t be able to provide effective tuition (eventually they may be even better than most human teachers), but because student motivation—a factor as significant as effective instruction—is something AI tutors will struggle to secure.

The situation is not without precedent. Intelligent tutoring systems—which adapt to the prior knowledge of the learner—have been around for decades, yet few have managed to find a widespread foothold in schools.

This is largely because personalised learning approaches fail to recognise just how important social norms, authority accountability, and human interaction are when it comes to motivation (and so learning). They have a motivation blindspot.

The ambient energy of ‘all the people around me learning’ is the biggest (and sometimes only) reason that many students learn. Motivation is a highly social phenomenon.

While some highly self-motivated or academically inclined students will undoubtedly benefit from personalised AI tools, relying solely on this model risks exacerbating existing inequalities. Those who already possess strong intrinsic motivation or robust support systems are best positioned to succeed, potentially further widening the disadvantage gap.

IMO, a more promising and equitable bet may be to use AI not to replace, but to enhance teaching (eg. as a planning thought partner) and even more powerfully: to enhance teacher development (eg. through training simulations or AI-enhanced coaching)... ideally in ways that emphasise the experience of the class as a collective, rather than isolate the student as an individual.

🎓 For more, check out this study on the effects of AI-enhanced human tutors.

Summary

• Personalised learning is experiencing an AI resurgence, but it is destined to flounder.

• Partly because it fails to recognise the important role of social motivation in learning.

• A more promising approach may be to use AI to enhance teachers and teacher expertise.

Little updates 🥕

• Review of RCTs on growth mindset in schools → finds strongest studies show near-zero impact, so large-scale investment may not be warranted.

• Classroom study on spelling practice in school → finds interleaving boosts long-term learning, especially for those with higher prior knowledge.

• Meta-analysis on arithmetic fluency → finds combining retrieval and strategy instruction best supports lasting fluency and deeper maths learning.

• Longitudinal study of ‘teaching to the test’ → finds increased TTT before exams may (in contrast to common views) actually boost intrinsic motivation, importance, and utility (HT Prof Hendy).

Get weekly 100-word research summaries & more → Learn about Snacks PRO