alt Hacker NewsI like the idea of "it's not about LLMs" better as a starting point. For example, someone could peel off from a shared coding session, ask an LLM to try a concept, then bring that LLM chat to the group to explore (and then yes close that window and go back to face-to-face). Make more efficient use of the face-to-face time.
You could be right but I can think of numerous frustrating code jams in my past when we burned a lot of precious face-to-face time on fussy setup or other fiddly stuff.
> You could be right but I can think of numerous frustrating code jams in my past when we burned a lot of precious face-to-face time on fussy setup or other fiddly stuff.
Agreed, LLMs are particularly good at this kind of task.
For example: my windowing system on Linux would intermittently freeze. Diagnosing it was a pain--so I bounced the logs off the LLM. It gave me a couple of hypotheses and the commands to enable the correct logs for when it happened again. After the third time it happened, it pinpointed that a particular USB hub was causing the issue. I removed the devices downstream from that hub and haven't had an issue since.