Do AI course tutors scaffold learning or smooth away the struggle?

Dr Jon Cardoso-Silva
📧 J.Cardoso-Silva@lse.ac.uk

10 Jun 2026

Are pedagogical AI tutors delivering?

Brave new words: how AI will revolutionise education, with Salman Khan ¡ DSI Event ¡ Wed 26 Jun 2024

📕 Read my book review.

  • Khanmigo was the best bet so far: the most funded tutor, Socratic by design, grounded in Khan Academy, backed by Gates and Microsoft.

  • At the DSI event in June 2024, Sal Khan called AI tutoring the biggest positive transformation education has ever seen. By April 2026 he told Chalkbeat that for many students it had been a non-event (so far?).

Dan Meyer, 15 Apr 2026 ¡ RIP Khanmigo & Edtech Industry Dreams of AI Tutors

What I’ve been building

  • The “Socratic” tutor is just the baseline. We gotta build on top of that.
  • The temporal dimension: we don’t need AI tutors that answer W02 questions the same way in W10. The tool must follow what should be currently active in the student’s learning journey.
  • I built one that moves with the teaching calendar: active skills, assignment phase, and what to withhold until it has been introduced.

DS105W (Winter 2025/26): formative practice W01–W04, then Mini Projects across the term.

When to route to what

When it works (1/2)

It gets Socratic when it’s needed

When it works (2/2)

But it does give direct guidance when it is called for

It reinforces what the course leader said

It aligns with the course leader’s intent for the course.
(The goal should be for “AI in the loop”, NOT “human-in-the-loop”)

But when it goes wrong… it’s VERY annoying

The answer can sound fluent and confident.
It makes you wonder, what other lines has this tool crossed?

The student should be the one justifying its choices. Here, the AI tutor offers a justification when the student didn’t even ask for it!

AI usage always increases near a deadline

DS105W (Data for Data Science) ¡ Winter 2025/26

  • 103 students enrolled
  • 83 consented to research
  • 47 shared at least one Claude tutor log

(I’m still going through the data)

Usage clustered around coursework deadlines.

Preliminary

Shared logs only; students may also have used ChatGPT or Claude outside the project.

What students asked for

Preliminary

  • Most messages were assignment guidance (red) or technical support (green)
  • The grey blocks are stretches with no protocol marker. I have not coded those yet
  • Next: GENIAL coding on those grey stretches in particular (why ‘No protocol’?)

What I do not know yet

  • I like to think my marking process is AI resilient, but submissions looked particularly strong this year.
    • Is it just performance? Or is it performance with learning?
    • Did students learn more because they had to engage more with the tool? Or did I build a fancy checklist?
  • GENIAL coding is next: grey no protocol stretches and just how influential was this tutor on my students’ thinking and work.

Thank you

Cardoso-Silva, J., Sallai, D., Kearney, C., Panero, F., & Barreto, M. E. (2025). Mapping Student-GenAI Interactions onto Experiential Learning: The GENIAL Framework. SSRN Electronic Journal. (Under Review)

Sallai, D., Cardoso-Silva, J., et al. (2024). Approach Generative AI Tools Proactively or Risk Bypassing the Learning Process in Higher Education. LSE Public Policy Review, 3(3), 7.

Blog series: jonjoncardoso.github.io/blog

📧 J.Cardoso-Silva@lse.ac.uk

Upcoming

Cardoso-Silva, J., & Sallai, D. (2026). Moving with the course: designing a chatbot tutor around the teaching calendar. (In preparation)