Writing, Reading, and Thinking in the Age of AI

Recent global surveys show widespread AI adoption among students and professionals, signaling a fundamental shift in how we engage with text. While some can already sense these changes in our daily practices, we are only beginning to understand their deeper implications. Mina Lee's research asks: How does AI influence what we write, how we write, and who we become as writers? And similarly, how does it influence our reading and critical thinking?

Writing in the Age of AI

Lee's research captures the process of human-AI collaboration to understand how AI changes our fundamental activities, focusing here on writing. In CoAuthor, she captured around 1,500 human-AI collaborative writing sessions and analyzed them to quantify AI's impact on language choice, idea generation, and collaboration patterns.

The CoAuthor findings and subsequent studies by multiple research groups reveal AI's dual impact on writers: while it can accelerate ideation and writing speed, it also risks homogenizing style, diminishing cultural nuance, and reducing writers' sense of ownership. These risks underscore the urgent need to carefully design AI writing assistants to harness benefits while mitigating potential harms.

Lee's recent Design Space project (a large collaboration with 36 researchers from 27 institutions analyzing 115 papers) maps the landscape of possibilities for AI writing assistants. We identified five interconnected aspects (task, user, technology, interaction, and ecosystem) with 35 dimensions and 143 codes, creating a comprehensive framework for researchers and designers to more effectively explore the design space of writing assistants.

Building on these insights, Lee's group designs targeted AI writing interventions and tests them through controlled experiments, examining their psychological effects, such as their potential to counteract human stereotypes.

Reading in the Age of AI

Her group’s research extends to examining how AI transforms reading comprehension. In an upcoming CHI 2026 paper, they examine AI's double-edged effects on literary interpretation. While AI assistance helps readers analyze poems (improving performance and narrowing gaps between novice and expert readers), their findings reveal a trade-off: AI enhances novices' experience while diminishing experts' experience. Furthermore, they observe that readers who heavily relied on AI assistance showed better performance on the task but lower pleasure. This result highlights the need to carefully calibrate AI assistance in cultural interpretation to preserve the experiences we value as humans.

Critical Thinking in the Age of AI

Lee's team also investigates AI’s impact on critical thinking. In another upcoming CHI 2026 paper, they use a synthesis essay writing task (i.e., reading and analyzing multiple documents and writing an argumentative essay based on them) to examine how the timing of AI access influences participants’ critical thinking task performance under different time availability. They found that AI access from the start improved performance under time pressure but impaired it when there was sufficient time, whereas starting without AI access showed the opposite pattern. These findings demonstrate that time constraints fundamentally shape whether AI augments or undermines critical thinking.

In another project, Lee's team explores UI/UX-based interventions, investigating how deliberate “frictions” (i.e., small, intentional barriers to AI use) can increase mindfulness, reduce overreliance, and improve task performance.

Across these projects, they systematically capture human-AI interaction and measure AI’s impact to better understand the evolving relationship between humans and AI. This knowledge informs the design of AI systems and interventions that amplify AI’s benefits while mitigating its risks, and deepens our understanding of what it means to write, read, and think in the age of AI.

Collaborations

Group members come from diverse backgrounds, including human-computer interaction (HCI), natural language processing (NLP), and computational social science (CSS), and collaborate broadly across disciplines, such as psychology, digital humanities, education, and media. They are always open to collaboration, so please reach out if you are interested.