Reintegrating the Human in AI-Augmented Learning

Why Aggregated Cultural Positioning Cannot Mediate Sociocultural Learning

by Zia Hassan

First, a few PhD jokes...

"Vygotsky walks into a bar. The bartender says 'What'll it be?' Vygotsky says 'I'll have whatever the person next to me is having—I'm in my ZPD for cocktails.'"

"Breaking: Freire starts a podcast. First episode title: 'Banking on Better Dialogue.'"

"Why did the CRT scholar bring a mirror to class? To reflect on institutional racism."

The Current Moment

AI is rapidly being integrated into education:

  • Automated feedback and grading
  • AI tutoring systems
  • Content generation
  • Even teaching entire courses

The Problem I'm Investigating

How does AI's aggregated cultural positioning fail to support the sociocultural learning processes that require specific, challengeable positioning?(Vygotsky, 1978; Gee, 2008; Ladson-Billings, 1995)

What this presentation will cover

  • Why AI is incompatible with authentic socioculturalism
  • The consequences/trajectory of our current moment
  • My proposed solution

Why is it...

AI chatbots can describe fasting during Ramadan... but has never experienced hunger

AI chatbots can synthesize every joke ever written... but it can't make you belly laugh

The Novel Joke Test - PhD Version

My thesis isn't late — it's just experiencing a delayed publication bias.

Why Does This Matter?

Humor is culturally situated (Flamson & Barrett, 2008)

Emerges from shared positioning, lived experience

AI can produce joke-shaped text

but cannot occupy the cultural position from which humor emerges

Brian Eno's "Mudge"

"AI output is like mudge—the color of water after painting with many colors" (Eno, 2024)

You get all of it, and none of it

How My Thinking Has Changed

AI is not void of cultural position

It has an "aggregated" cultural position

Shout out to KM

...which is worse than having none

Part 1

Does AI Have a Cultural Position?

I used to say that it doesn't have a sociocultural position. However...

It turns out that, yes, AI is Trained on culturally situated data (Bender et al., 2023)

Every training corpus reflects:

  • Historical moment
  • Language patterns
  • Cultural practices
  • Power structures

The more common patterns (which are centered around the dominant western culture) are weighted and therefore more likely to show up as output (Fenech-Borg et al., 2024; Bender et al., 2023).

This is not great, because...

LLMs "position" is just an average of its entire biased corpus

It cannot be challenged or changed through dialogue

What Sociocultural Learning Requires

Vygotsky: Mediation & Development (1980)

Learning occurs through:

  • Culturally shaped tools
  • Interaction with more knowledgeable others
  • Zone of proximal development

Importantly...

The "other" must occupy a specific cultural position you can engage with (Freire, 1970)

Gee: Discourse & Identity

  • Learning means acquiring Discourses
  • AI chatbots can mimmick discourse; but it is the average discourse

Fine-tuning isn't the solution

Dewey see the problem?

sorry

Let's talk Dewey

Claude has never had its heart broken.

ChatGPT has never tasted ice cream in July.

Gemini has never been a bored 10 year old on summer break.

Dewey's Qualitative Thought (1931)

Dimension Human Learning (Dewey) AI "Learning"
Experience Qualitative, felt, immediate - "total seizure" of wholeness Statistical pattern recognition - no felt quality
Modification Each experience modifies the learner - continuous growth Fixed weights after training - no modification through use
Embodiment Body is site of experience - sensing, feeling, doing No body, no sensory experience
Social Context Learning through interaction with positioned others Aggregates patterns from text about interactions
Purpose Learner has stakes, desires, intentions No purposes - responds to prompts without intention
Integration Experience forms unified whole - "an experience" Outputs are stitched-together patterns - no unity
What Changes Habits, dispositions, ways of being - self transforms Weights, parameters - no self to transform

Part 2: A Few Cascading Problems

What happens when aggregated positioning mediates learning?

Problem 1: The Reproduction Mechanism (Bourdieu, 1970)

AI's aggregated position appears neutral

But it actually encodes dominant patterns

Students without prior capital are at risk; and it is insidious

Problem 2: The Progressive Veneer

AI marketed as equity solution

But lacks the very thing that interrupts reproduction: relational friction with positioned humans (Hassan, 2025)

Problem 3: Mechanized Creativity

AI produces what Runco (2014) would call "little-c" creativity

And violates Plucker's (2004) definition of sociocultural creativity

Who Thrives with AI Mediation

Students who already have:

  • Cultural capital to decode expectations (Bourdieu, 1986)
  • Ability to navigate dominant discourse (Gee, 2015)
  • Internalized academic identity (Bourdieu, 1990)
  • Prior institutional knowledge (Lareau, 2011)

Who Loses

Students who rely on:

  • Teachers to mediate cultural difference (Ladson-Billings, 1995)
  • Relational support to navigate institutions (Noddings, 2005)
  • Seeing positioned models who succeeded (Steele & Aronson, 1995; Ladson-Billings, 1994)
  • Learning to decode unstated expectations (Delpit, 2006)

Part 3:

The Sociocultural Reintegration Framework

The Core Question

Which educational activities require specific cultural positioning?

Which can work with aggregated positioning?

How do we design for this?

Three Dimensions

  1. Culturally Positioned Feedback
  2. Relational Friction
  3. Positioned Dialogue & Consciousness-Raising

Dimension 1: Culturally Positioned Feedback

Problem: AI responds from aggregated position, cannot engage cultural difference

Reintegration: Feedback from positioned subjects who can negotiate

Dimension 2: Relational Friction

Problem: AI understands the shape of concepts, but has never experienced life

What students need:

  • Exchange of ideas with entities who can actually have positions and take responsibility
  • Learning embodied practice, not just product
  • Engaging with the friction necessary for learning and growth

Dimension 3: Positioned Dialogue & Consciousness-Raising

Problem: AI has aggregated positioning and no consciousness - cannot engage in transformative dialogue

Reintegration: Students need dialogue with positioned humans who can:

  • Mediate between cultural practices
  • Be challenged and transformed through encounter
  • Negotiate meaning and power together

Take aways

  • Figure out where positioning matters and protect it
  • Socioculturalism in education is at risk of becoming eroded in the name of "efficiency"
    • Don't believe the hype
  • Friction≠Inefficiency

References

Bender, E. M., Gebru, T., McMillan-Major, A., & Shmitchell, S. (2021). On the Dangers of Stochastic Parrots: Can Language Models Be Too Big? 🦜. Proceedings of the 2021 ACM Conference on Fairness, Accountability, and Transparency, FAccT ’21, 610–623. https://doi.org/10.1145/3442188.3445922

Bourdieu, P. (2018). Cultural Reproduction and Social Reproduction. In R. Brown (Ed.), Knowledge, Education, and Cultural Change (1st ed., pp. 71–112). Routledge. https://doi.org/10.4324/9781351018142-3

Delpit, L. (2006). Other people’s children: Cultural conflict in the classroom. the New press.

Fenech-Borg, E. Z., Meznaric-Kos, T. P., Lekovic-Bojovic, M. D., & Hentze-Djurhuus, A. J. (2025). The Cultural Gene of Large Language Models: A Study on the Impact of Cross-Corpus Training on Model Values and Biases (arXiv:2508.12411). arXiv. https://doi.org/10.48550/arXiv.2508.12411

Flamson, T., & Barrett, H. C. (2013). Encrypted humor and social networks in rural Brazil. Evolution and Human Behavior, 34(4), 305–313. https://doi.org/10.1016/j.evolhumbehav.2013.04.006

Freire, P. (1970). Freire, P. (1970). Pedagogy of the Oppressed. New York Seabury Press. - References—Scientific Research Publishing. https://www.scirp.org/reference/referencespapers?referenceid=1415936

Gee, J. P. (n.d.). A Situated Sociocultural Approach to Literacy and Technology.

Hassan, Z. (2026). Friction and Coordination in Schooling:

Ladson-Billings, G. (1995). Toward a Theory of Culturally Relevant Pedagogy. https://doi.org/10.3102/00028312032003465

Lareau, A. (2011). Unequal childhoods: Class, race, and family life (2nd ed., with an update a decade later). University of California Press.

Noddings, N. (2005). The challenge to care in schools: An alternative approach to education (Second edition). Teachers college press.

Plucker, J. A., Beghetto, R. A., & Dow, G. T. (2004). Why Isn’t Creativity More Important to Educational Psychologists? Potentials, Pitfalls, and Future Directions in Creativity Research. Educational Psychologist, 39(2), 83–96. https://doi.org/10.1207/s15326985ep3902_1

Runco, M. A. (2014). CREATIVITY: THEORIES AND THEMES RESEARCH, DEVELOPMENT, AND PRACTICE (Second edition). ELSEVIER.

Steele, C. M., & Aronson, J. (1995). Stereotype threat and the intellectual test performance of African Americans. Journal of Personality and Social Psychology, 69(5), 797–811. (1996-12938-001). https://doi.org/10.1037/0022-3514.69.5.797

Vygotsky, L. S., Cole, M., John-Steiner, V., Scribner, S., & Souberman, E. (2012). Mind in society: Development of higher psychological processes. Harvard University Press. https://books.google.com/books?id=u2PP6b0ddtoC

Slides available at http://ziahassan.com/sociocultural