Building systems that enable self-learning of skills

Building systems that enable self-learning of skills

MIT Open Learning

Q&A with Dishita Turakhia, PhD candidate at MIT CSAIL

Dishita Turakhia is currently a 5th year Ph.D. candidate in the MIT Electrical Engineering & Computer Science (EECS) department at MIT. She conducts research in the field of Human-Computer Interaction in the HCIE group in MIT Computer Science and Artificial Intelligence Lab (CSAIL). She and professor Stefanie Mueller have been recipients of two past education grants: Improving learning with adaptive physical tools and Learning effectiveness of dynamically generated tutorials. Turakhia and Mueller have also recently been granted a third award studying the role of personalized reflection in the learning of maker-skills.

Turakhia is also a recipient of the Meta (Facebook) Ph.D. Research Fellowship (2022) and the MIT Edwin S. Webster Graduate Fellowship (2018). In her research, she builds systems for self learning of skills, such as motor skills, fabrication skills, and maker skills. While her current research lies at the intersection of system design and learning sciences.

What initially drew you to study human-computer interaction, and in particular the effects on education?
My background is in architecture, where human needs and behavior are at the center of the design process. So after completing my dual master’s in computational design (architecture) and computer science at MIT, transitioning into the Ph.D. program in human-computer interaction (HCI) seemed like an exciting next step. This is because HCI research is highly interdisciplinary and allows me to bring my knowledge and skills in computational design to build systems for enriching the human experience, especially the experience of skill-learning.

I have always been deeply interested in studying how humans learn and how learning environments deeply impact our skill development. Most of us have that one teacher who altered our life path because their unique way of teaching sparked an interest in a topic that defined our careers. Education is a great equalizer, but unfortunately, not everyone has access to good quality education. This realization inspired my research in building systems that enable self-learning of skills. If we can build systems that enhance the experience of self-learning of skills while also making it accessible, we might get one step closer to unlocking the full potential and passions of people.

What role do you see for computer science and artificial intelligence in human learning?
Two areas where Computer Science and AI can play a significant role in improving the human learning experience are (i) in expanding access to learning and (ii) in personalizing the learning environments.

Our educational workforce is frequently overworked and understaffed and the shortage of teachers and human resources in education continues to be a challenge. While we can never replace educators and teachers, we can certainly build systems that can complement and support our educators so that high quality and personalized education can be made accessible to learners from around the world. With the advancement in enabling technologies such as sensing, computer vision, and AI and ML algorithms, our systems are capable of a more sophisticated understanding of the learner’s skill levels, progress, preferences, and even capabilities. This capability of computational systems allows us to design a personalized learning plan for every learner based on their skill levels so they can learn at their own pace, a feat that can be labor-intensive otherwise. When required, teachers can intervene and provide additional scaffolding and in-person teaching.

While self-learning systems and platforms are widely available for learning subjects like sciences, math and programming, they are relatively unexplored for physical skills, such as motor and maker skills,, making this an exciting research topic!

Tell us a little about your work with personalized reflection for MITili.
The inspiration for this project comes from the seminal work titled “The Reflective Practitioner” by philosopher and learning theorist, Donal Schon, who was also a professor of Urban Studies at MIT. His work underscored the role of deliberate reflection and inquiry in learning and led to a transformation in design and architecture education. Architecture studios are unique learning environments where consciously reflecting on one’s work through a dialogue either internally, with their peers, or their educators is a crucial aspect of a learner’s journey. Even in my own education, a lot of my learning in architecture and other subjects occurred through engaging in a reflective dialogue. So bringing this aspect of reflection into my core research and how I design learning systems is, in a way, a personal pursuit.

Through this project with MITili, my team and I are exploring how reflection can contribute to the learning of maker skills, such as learning electronics, digital fabrication, etc. Maker skills provide a unique testing bed for studying reflection because of the open-ended nature of the maker activities. We are building a system that prompts the learner to reflect on their activities at different stages of their project. So for example, imagine that a learner is working on the assignment of designing a novel interactive smartwatch. Our system engages in a reflective dialogue with the learner by first monitoring their making activity, identifying the right instances to prompt a reflection task, and then recording their reflections. There are several levels of reflective prompts that go from being highly specific, such as the correct use of tools and components, to being highly open-ended, such as the overall design of the smartwatch. Our goal is to study how reflection impacts the learner’s experience and learning trajectory and to identify which stages of the project are most suitable for and benefit from reflection. We plan to run controlled trials to study the learning gains from using our system and our findings will inform the design of learning systems in the future.

We know that education technology isn’t a silver bullet for struggling learners. What have you seen in your research and maybe even on campus that gives you hope that human-computer interaction can help make learning more effective?
There is ample evidence to suggest that learning is enhanced when humans are at the center of the process, which is precisely what the research in the field of human-computer interaction is trying to achieve.

Often, the approach in the design of systems is techno-centric, that is, identifying an exciting technology and then identifying new applications using that technology. Unfortunately, this approach usually sidelines the user, and in our case, the learner. By bringing the learner back to the center of the design of our learning systems, much of our educational technology can be revolutionized for the better. For example, in both my prior projects (also funded by MITili) — adaptive learning of motor skills, and game-based learning of fabrication skills — we designed the systems using off-the-shelf computer vision algorithms and applied them in a way that personalized the learning experience. In the adaptive learning project, this personalization was implemented in how quickly or slowly the task difficulty of training adapted for the learner, and in the game-based learning project, the personalization was implemented in the choice of games, characters, and the game object for fabrication assigned to the learner. Our studies showed that this approach to learning not only improved the learning gains but also improved engagement and motivation among learners.

And speaking of engagement and motivation, what better place to witness how these factors can significantly improve effective learning than right here at MIT? During my time here at MIT, classes where I have learned the subject most effectively, are those where the professors have taught the content in several different ways to make it explainable and accessible to a class with diverse students with varied backgrounds or expertise.

What is your favorite thing about MIT?
MIT is an endlessly inspiring place. Since I came to here in 2015, I have been starry-eyed and in awe of how incredibly gifted and creative everyone is around me. Over the years, I have learned how the environment here shapes us to be incessantly hardworking and resilient. As daunting as being at MIT can be, being around such a unique group of passionate researchers and students also keeps me constantly inspired to push the boundaries of my work.

Originally published at on September 14, 2022.

Building systems that enable self-learning of skills was originally published in MIT Open Learning on Medium, where people are continuing the conversation by highlighting and responding to this story.


Open Learning newsletter