Uploaded January 2025 | Updated September 2026, 1 day ago
Abstract:
As generative AI reshapes the future of work by automating a wide range of productivity tasks, it also introduces significant challenges. In the short term, users face difficulties interacting with AI using natural language and verifying AI-generated outputs. Over the long term, there are concerns about over-reliance on AI, which can lead to skill degradation and negative impacts on human cognition. This talk addresses these challenges and explores potential solutions by focusing on *programming*–a productivity task transformed by Large Language Models (LLMs). I will begin by examining the cognitive impacts of AI-assisted programming, presenting findings from a study with novice programmers. This study investigates whether access to AI fosters independent problem-solving skills or leads to dependency and skill degradation. Building on these insights, I will present a systematic design exploration of seven interfaces and interventions aimed at cognitively engaging programmers with AI-generated code to mitigate risks of over-reliance. I will demonstrate how involving users in the AI's step-by-step problem-solving process significantly increases metacognitive self-awareness without introducing excessive friction. Shifting focus to productivity, I will introduce a novel human-AI interaction technique in which the LLM's reasoning and problem-solving process is displayed progressively using structured, editable UI elements. I will demonstrate the effectiveness of this technique in the context of AI-assisted data analysis, showing how it enhances user control and verification capabilities.
Bio:
I'm Majeed, a PhD candidate in Computer Science at the University of Toronto, where I've been fortunate to be advised by Prof. Tovi Grossman. My research in Human-Computer Interaction focuses on addressing fundamental challenges surrounding interaction and cognition within the evolving landscape of programming with generative AI.
My work includes (a) studying the implications of AI on over-reliance when learning to code and, (b) designing novel interfaces and interventions that cognitively engage programmers with AI-generated solutions, (c) developing human-AI interactions that involve users in editing the AI's chain-of-thought reasoning, and (d) deploying pedagogical AI assistants that promote independent problem-solving.
Prior to my PhD, I earned a master's degree in Computer Science from the University of Maryland, where I worked with Prof. Jon Froehlich at the Makeability Lab and the HCIL. During that time, I led the two-year design, development, and evaluation of MakerWear—a tangible electronic construction kit that enables young children in creating their own interactive wearables. This work received a Best Paper Award at CHI 2017 and a Best Late Breaking Work Award at CHI 2016.
I have been fortunate to collaborate with leading researchers in academia and industry, including two internships at Microsoft Research, where I was hosted by Dr. Jack Williams in 2023 and Dr. Rob DeLine in 2017. Additionally, I was a visiting research scholar at UC Berkeley, hosted by Prof. Bjoern Hartmann in 2017.
To date, I have published 10 full conference research papers in top-tier HCI venues, including CHI, UIST, IUI, SIGCSE, and IDC. My recent work in human-AI interaction is among the most highly cited within the HCI and Computer Science Education Research community.
Abstract:
As generative AI reshapes the future of work by automating a wide range of productivity tasks, it also introduces significant challenges. In the short term, users face difficulties interacting with AI using natural language and verifying AI-generated outputs. Over the long term, there are concerns about over-reliance on AI, which can lead to skill degradation and negative impacts on human cognition. This talk addresses these challenges and explores potential solutions by focusing on *programming*–a productivity task transformed by Large Language Models (LLMs). I will begin by examining the cognitive impacts of AI-assisted programming, presenting findings from a study with novice programmers. This study investigates whether access to AI fosters independent problem-solving skills or leads to dependency and skill degradation. Building on these insights, I will present a systematic design exploration of seven interfaces and interventions aimed at cognitively engaging programmers with AI-generated code to mitigate risks of over-reliance. I will demonstrate how involving users in the AI's step-by-step problem-solving process significantly increases metacognitive self-awareness without introducing excessive friction. Shifting focus to productivity, I will introduce a novel human-AI interaction technique in which the LLM's reasoning and problem-solving process is displayed progressively using structured, editable UI elements. I will demonstrate the effectiveness of this technique in the context of AI-assisted data analysis, showing how it enhances user control and verification capabilities.
Bio:
I'm Majeed, a PhD candidate in Computer Science at the University of Toronto, where I've been fortunate to be advised by Prof. Tovi Grossman. My research in Human-Computer Interaction focuses on addressing fundamental challenges surrounding interaction and cognition within the evolving landscape of programming with generative AI.
My work includes (a) studying the implications of AI on over-reliance when learning to code and, (b) designing novel interfaces and interventions that cognitively engage programmers with AI-generated solutions, (c) developing human-AI interactions that involve users in editing the AI's chain-of-thought reasoning, and (d) deploying pedagogical AI assistants that promote independent problem-solving.
Prior to my PhD, I earned a master's degree in Computer Science from the University of Maryland, where I worked with Prof. Jon Froehlich at the Makeability Lab and the HCIL. During that time, I led the two-year design, development, and evaluation of MakerWear—a tangible electronic construction kit that enables young children in creating their own interactive wearables. This work received a Best Paper Award at CHI 2017 and a Best Late Breaking Work Award at CHI 2016.
I have been fortunate to collaborate with leading researchers in academia and industry, including two internships at Microsoft Research, where I was hosted by Dr. Jack Williams in 2023 and Dr. Rob DeLine in 2017. Additionally, I was a visiting research scholar at UC Berkeley, hosted by Prof. Bjoern Hartmann in 2017.
To date, I have published 10 full conference research papers in top-tier HCI venues, including CHI, UIST, IUI, SIGCSE, and IDC. My recent work in human-AI interaction is among the most highly cited within the HCI and Computer Science Education Research community.
