I haven’t had the chance to test out this specific feature yet so I can’t speak to it specifically, but I’ve found it extremely handy being able to “interrogate” the contents of a web page using an LLM to specifically summarize whatever particular details I’m after from it. I still miss Firefox’s “Orbit” extension for how it could summarize the content of Youtube videos, that was great for situations where someone would link an hour-long video to back up some argument they were making. Or just generally deciding whether it was worth watching an hour-long video, or skipping to whatever the basic point of the video was without all the irrelevant cruft.

In another thread on this subject I mentioned how I’ve lately found that Copilot is quite handy at creating quick Tampermonkey scripts to do tasks on specific web pages. It’d be nice if that was integrated directly into the browser, so that I could ask it for a script to do a task and it would have access to the page’s source as context while creating it.

Current AI features are fairly simple, but I think the technology’s got plenty of neat new applications coming in the near future. It frustrates me how many people want those applications to be limited to corporate-controlled browsers like Edge or Chrome.