You can also just skim it. It feels like LLM summarization boils down to an argument to substitute technology for media literacy.

Plus, the latency on current APIs is often on the order of seconds, on top of whatever the page load time is. We know from decades [0] of research that users don't wait seconds.

[0] https://research.google/blog/speed-matters/

It makes a big difference when the query runs in a sidebar without closing the tab, opening a new one, or otherwise distracting your attention.