Our Vision

One of the ways that humans care for each other and contribute to each other’s well-being is through thinking about each other. So instead of generating content for people, AI should help people think better and more deeply about the people they’re serving---whether readers, students, teammates, or communities.

The Challenge

Naive uses of generative AI, focused narrowly on reductionist notions of efficiency and productivity, threaten human thought by promoting cognitive disengagement and shortcut thinking.

Writing is one of the main uses of generative AI today. But most uses cede substantial parts of the writing process to LLMs. But writing is thinking, not just content production.

Although people could carefully read and verify LLM output, the cognitive shortcut is extremely tempting. And even when we do read the output carefully, we don’t even know what we could have thought about had we engaged more deeply with the writing process.

Our Work

Our work asks questions like:

• What is the minimal essence that the computer could helpfully contribute to our thinking that’s *useful *(gets us to think of something we wouldn’t have thought of on our own) and not invasive (stepping on our own thoughts/agency, cognitive interference, …)?
• What processes (of reflective revision, etc.) could we encourage through making them “default” in writing tools?
• What collaboration patterns become possible when humans craft computationally mediated reflective practices together?

But AI could help us think better about others: expressing thoughts in ways that are easy to correctly understand, anticipating the otherwise “unanticipated” outcomes of our words and actions, and encapsulating practices of reflection and mindful action.

Want to learn more? Meet our team or explore our projects and publications.