In thinking about the future of large context window models, how do we evaluate whether changes user behavior in unexpected ways — and how might this influence future innovation?

[sabastion]

In thinking about the future of large context window models, how do we evaluate whether changes user behavior in unexpected ways — and how might this influence future innovation?