LLMs, Critical Thinking, and the Wrong Question
A closer look at what studies of AI-assisted writing can—and cannot—tell us about cognition, effort, and the quality of the work people produce.
Writing Arguments in
Arguments in
development.
Essays from the overlap of learning science, measurement, data, and system design. The queue is visible because unfinished questions are often more useful than polished certainty.
Dispatch queue Ideas in development
Research Reports Should Be Written with LLMs
The argument is not that models should decide what the evidence means. It is that researchers should spend less time manufacturing prose and more time improving the analysis behind it.