AI in Edu

"Every child will have an AI tutor that is infinitely patient, infinitely compassionate, infinitely knowledgeable, infinitely helpful. The AI tutor will be by each child’s side every step of their development, helping them maximize their potential with the machine version of infinite love."

- Marc Andreessen

Ongoing GenAI Research Projects & Contributions in Learning Sciences

April 2025

Gen AI-powered Dynamic Peda-Agents for Online Learning

Visual showing the systematic process involved in the development of GenAI-powered learning companion.

The educational implications of the dynamic peda-agent as study buddy are astounding.

This visual outlines my multi-step pipeline for designing AI-powered dynamic Peda-Agents, also referred to as synthetic agents—which are animated digital tutors that support online [math] learning in emotionally intelligent and pedagogically meaningful ways. Starting with visual character creation (via any text-to-image models) and progressing through image-to-video generation, natural speech synthesis, and real-time audio-visual lip syncing (using tools such as WAN 2.1, OpenAI API, LatentSync), each step is carefully calibrated for educational effectiveness. In the final stage, the MathSpring (MS) research team at UMass Amherst conducts a human-in-the-loop review of the synthetic agent to ensure accuracy, tone, and alignment with instructional goals and MS teaching philosophy. This approach reflects our ongoing research to develop emotionally responsive, story-driven, and culturally aware math learning companions.

This work builds on my July 2023 original video demonstrating how to integrate generative AI with intelligent tutoring systems—and marks a pivotal step toward reimagining the future of learning companions as emotionally attuned, pedagogically rich, and culturally resonant guides in math education.

Image credits: Sai Gattupalli

Expertise related to UN SDGs

In 2015, UN member states agreed to 17 global Sustainable Development Goals (SDGs) to end poverty, protect the planet and ensure prosperity for all.

My research contributes towards the #4 Quality Education, #8 Decent Work and Economic Growth, and #17 Global Partnership for Sustainable Development.

Perspective Article: Educators + AI super power = Super Educators

April 2025

See you at ICALT 2025!

I'm excited to announce that my latest research on On-Device AI for Computing Education has been accepted for presentation at the IEEE International Conference on Advanced Learning Technologies (ICALT) 2025. This research introduces SHIELD, a novel framework for using on-device AI to expand access to computing education in contexts with limited or no internet connectivity. By centering equity, data privacy, and culturally responsive pedagogy, this work explores how offline AI systems can empower underserved learners and reshape the digital divide in K–12 education.

This work adds to latest research on AI Literacy, low-/no-infra schools, equitable and advanced learning technologies. More details to come as we approach the conference.

Full schedule to be announced: https://tc.computer.org/tclt/icalt-2025/

Perspective Article: What comes next, nobody knows

October 2024

Come to my educator poster session at MassCUE Fall 2024

Thanks to everyone who attended my Educator Poster Session at the MassCUE Fall 2024 Conference.

Discover on-device AI wrappers, implication for educators, and application in teaching and learning contexts were discussed and demonstrated. For a deeper look into these innovative and equitable education technologies, see full poster:

https://doi.org/10.7275/g6qd-km54

Prompt Literacy bridges Human Creativity and AI in Modern Education

Read our research article titled "Prompt Literacy: A Pivotal Educational Skill in the Age of AI," where we explore the critical role of effective prompting in achieving desired outcomes from foundation models. We emphasize how mastering prompt literacy is essential for both educators and learners to harness the full potential of AI technologies in education.

This work is a collaborative effort with colleagues Dr. Robert Maloy and Dr. Sharon Edwards from the College of Education, University of Massachusetts Amherst.

Cite our work (APA):

Gattupalli, S., Maloy, R. W., & Edwards, S. A. (2023). Prompt literacy: A pivotal educational skill in the age of AI. https://doi.org/10.7275/3498-wx48

Are you a K-12 Math Teacher or Tutor? We made a math readability chatbot for you!

Discover Estella Explainer Math Bot 2 (EEMB 2), a GPT-4 powered chatbot that simplifies math word problems to enhance student understanding and engagement.

Chat with the Estella Explainer Math Bot 2 here. Or click anywhere on the image above.

Our free chatbot is designed to assist elementary math teachers and tutors worldwide teaching math topics in English. Powered by ChatGPT, our AI tool offers personalized readability support for math word problems, provides valuable math resources, and introduces interactive problem-solving techniques. Our goal is to enhance both teachers' and students' understanding and engagement in mathematics across formal and informal learning environments, as it helps educators (pre-service and in-service) improve their programs and practices.

In our internal testing, the EEMB 2 chatbot has consistently outperformed our expectations. While we are continuously making improvements to the chatbot, you may sometimes experience dissatisfaction with the performance of the bot. I encourage EEMB 2 users to keep refining your prompts and try again. If you have any feedback, I would love to hear: sgattupalli at umass dot edu.

A free ChatGPT account is required. While we do not collect any user inputs, ChatGPT does, and uses it for their model training purposes. Users must be at least 13 years old to use ChatGPT.

Technicals of EEMB 2:

Aligned with Usable Math's teaching philosophy.

10-shot learning for fine-tuning, with prior Massachusetts Comprehensive Assessment System (MCAS) test items with human-crafted hints.

Image source: Unsplash.

Employs Flesch–Kincaid readability metric for math explanations. See decision tree on our blog.

Image source: Wiki.

Prompting instructions based on Universal Design for Learning (UDL) principles.

Image source: Unsplash.

I developed EEMB 2 in collaboration with math education faculty:

Sharon A. Edwards, EdD, & Robert Maloy, EdD.

See how we are use generative AI technologies in elementary math teaching and learning contexts.

Screenshot showing Usable Math's AI-Enhanced Learning page. Click on the image, or click here to visit.

Encouraging Ethical & Responsible AI Use in College Writing

In my College Writing course (ENGLWRIT 112) at UMass Amherst, I aim to equip my students with the skills to use generative AI tools such as ChatGPT and Claude responsibly and ethically. By emphasizing prompt literacy and ethical AI usage, I guide my students to critically engage with these tools, ensuring that AI enhances, rather than replaces, their creative and analytical efforts. This approach fosters a balance between AI’s capabilities and the essential human skills of critical thinking and writing.

Read current AI usage policy (FA 2024). Previous iteration available on Lance Eaton's Syllabi Policies for AI Generative Tools. More on my blog article.

Perspective Article: AI Literacy is fundamental for hybrid intelligence

LLMs as Hint Generators in Online Math Learning

Honored with presentation in AIED 2023

In this study, we explore pre-service teachers' perceptions of large language model (LLM)-generated hints in the context of online mathematics education. This research focuses on the potential of LLMs, such as GPT4, to provide walkthroughs and guidance in math problem-solving. While human-generated content, especially visual aids, remains preferred, LLM-generated hints show promise in enhancing the teaching process. Presented at AIED 2023, this paper highlights both the opportunities and challenges of integrating AI into math education and suggests directions for future development.

Cite our work (APA):

Gattupalli, S. S., Lee, W., Allessio, D., Crabtree, D., Arroyo, I., Woolf, B. P., & Woolf, B. (2023). Exploring Pre-Service Teachers' Perceptions of Large Language Models-Generated Hints in Online Mathematics Learning. In LLM@ AIED (pp. 151-162). https://ceur-ws.org/Vol-3487/paper10.pdf.

In collaboration with:

Will Lee, Danielle Allessio, Danielle Crabtree, Ivon Arroyo, & Beverly Woolf.

Banner Image Reference:

Cukurova, M. (2024). The interplay of learning, analytics and artificial intelligence in education: A vision for hybrid intelligence. British Journal of Educational Technology. https://doi.org/10.1111/bjet.13514.