Eye on A.I. · Tuesday, September 8, 2026
Alexander Whedon points out that current AI models struggle with context window limitations, which is a primary reason they spend so much time reading code. If models could process larger amounts of code at once, their efficiency in understanding and solving problems would significantly increase.
“This is a limitation of the current models, right? So you can't just pass in the entire codebase to the model. You have to chunk it up, you have to summarize it, you have to find the relevant parts. And that's why the model spends so much time reading.”