1 of 19

AI Tutoring and Self-Regulation in Schools: A Pilot Study 

Dr. Fátima Díaz Freire

Special Interest Group

16 of April 2026

LEAD Graduate School & Research Network

2 of 19

What I want to discuss in this SIG today

2 | Fátima Díaz Freire

  • My view about the impact of AI in education
  • A show&trial of the LLM designed using MITCA
  • Open debate
    • Your opinions and views on the topic

AI will not determine the future of education on its own, design will

Olga Viberg – LEAD Retreat 2026

(feedback!)

3 of 19

AI Impact on students learning

3 | Fátima Díaz Freire

As AI tools can resolve tasks and homework faster and more efficient than humans, students don’t feel the need to do it by themselves

memorize content only to reproduce it in a test and then forget it

copy-paste info without reviewing it

Hallucinations

Is this a new problem?

No, it has been happening for years

AI-produced essays

(=no learning nor understanding)

GEN-AI just made the need for self-regulation more evident

false information and ”fake news”

4 of 19

Guided learning mode

4 | Fátima Díaz Freire

When unsupervised, why use that instead of:

fast

direct

correct answer?

    • if you have the opportunity to do things faster, why wouldnt you?

cognitive off-load:

5 of 19

Cognitive off-loading

5 | Fátima Díaz Freire

Kosmyna et al. (2025): Your Brain on ChatGPT: Accumulation of Cognitive Debt when Using an AI Assistant for Essay Writing Task

https://arxiv.org/abs/2506.08872

54 participants writing easys (Kosmyna et al. 2025)

LLM assistance

search engine use

or no tools at all

Heavy & substitutive use of AI for writing may come with cognitive costs

6 of 19

Some (early) evidence about AI and learning

6 | Fátima Díaz Freire

Gerlich, M. (2025). AI tools in society: Impacts on cognitive offloading and the future of critical thinking. Societies15(1), 6.

Zhi, J., Kumar, H., & Lee, M. (2026). Investigating the Effects of LLM Use on Critical Thinking Under Time Constraints: Access Timing and Time Availability. arXiv preprint arXiv:2603.08849.

use AI

- critical-thinking scores

trust in AI

+ cognitive offloading

(Gerlich 2025)

-

+

early LLM

continuous LLM

No LLM

late LLM

+ essay performance

+ critical thinking

+ recall/internalization

10 min

30 min

(Zhi et al., 2026)

7 of 19

7 | Fátima María Díaz Freire

8 of 19

MITCA (MY RESEARCH!)

8 | Fátima María Díaz Freire

SOME CONTEXT AND CURRENT PROJECT :)

9 of 19

9 | Fátima Díaz Freire

Cyclic model of self-regulation phases (Zimmerman & Moylan, 2009)

Self-reflection phase

Self-Judgment

Self-evaluation

Causal attribution

Self-Reaction

Self-satisfaction/affect

Adaptive/defensive

Forethought phase

Task analysis

Goal setting, strategic planning

Self-motivation beliefs

Self-efficacy, outcome expectations,

task interest, goal orientation

Performance phase

Self-control

Task strategies, self-instruction, time management, help-seeking.

Self-observation

Metacognitive monitoring, self-recording

MITCA

Process preceding doing homework

Step 1: Specific

Step 2: Varied

MITCA

Process related to doing homework

Step 4: Weekly

MITCA

Once tasks are completed

Step 5: Evaluated

Step 3: Worthwile

10 of 19

10 | Fátima Díaz Freire

Describing the mental work and content, not just the activity number or textbook page.

Including exercises that involve developing, organizing, or reviewing information.

Communicating the utility, interest, and applicability of the task (e.g., What is the purpose of doing this exercise?).

Organized throughout the week, promoting time management and autonomy.

Feedback, whether individual or group-based, highlighting positive aspects and areas for improvement.

E.g.: Instead of "exercise 1 on page 21 of the Math textbook," it would be "Solve percentage problems."

For the previous example, it could be:

"It will help you better understand how discounts work when making a purchase."

Diverse

Specific

Valuable

Plannified

Evaluated

MITCA: a quality homework proposal

11 of 19

LLM based on MITCA principles

11 | Fátima Díaz Freire

12 of 19

LLM based on MITCA principles

12 | Fátima Díaz Freire

+

Combined use of LLM and template

13 of 19

Some characteristics

13 | Fátima Díaz Freire

  • Model: gemma-4-31b-it (last versión by Google)
  • Working on local (for now)

Response principles:

Respond in clear English (near future: German).

Keep answers concise, practical, and chat-friendly.

Use the provided context: MITCA step, topic, task, objective, due date.

Do not provide final answers, completed homework, or ready-to-submit text.

Do not solve the learner's exact assignment for them, even if they ask directly.

Ask the learner to attempt the next step, then help them check or improve it

14 of 19

MITCA steps LLM

14 | Fátima Díaz Freire

T (Planning): 

choose strategy, split into steps, estimate time, gather resources.

V (Value): 

connect usefulness, relevance, and personal meaning.

P (Activation): 

activate prior knowledge and detect gaps.

R (Reflection): 

evaluate process, extract lessons, transfer learning

M (Motivation): 

sustain focus, effort, and self-efficacy.

S (Supervision)

monitor understanding/progress and correct blocks.

Forethought phase

Working phase

Reflection phase

15 of 19

LET ME SHOW YOU A LITTLE BIT :)

15 | Fátima María Díaz Freire

16 of 19

MITCA-AI Pilot Study

16 | Fátima Díaz Freire

G1 – Group 1

MITCA Method �(No LLM)

G2 – Group 2

MITCA LLM

Students (13-15 y.o) will work for 5 weeks on tasks in the clasroom

Train highschool teachers on the MITCA Method

Pretest

(classroom set)

5 weeks / 40 min per week

Postest

Looking for 1 high school &

3-4 teachers!

🎉

17 of 19

Measurements

17 | Fátima Díaz Freire

Teachers

Interviews with teachers

Are the prompts of MITCA – AI useful?

�What did they find the most useful? What was the least?

Impressions about AI and education

Students

Self-regulation of Learning

Motivated Strategies for Learning Questionnaire (MSLQ) that has been validated for adolescent samples (Morais et al., 2025)

Critical Thinking

Critical Thinking Disposition scale (CTD) (Liu et al., 2026; Sosu, 2013).

18 of 19

18 | Fátima Díaz Freire

Google forms Feedback

Your opinion is very helpful!!

19 of 19

19 | Fátima María Díaz Freire

Thank you!

Gracias!

Vielen Dank

Contact and personal information