Interesting 🧵

Show Post

Use the # in front of the name to get more exposure. #SpaceX

Show Post

Perhaps for current-event reporting, you'd want the output tightly grounded in both contemporary sources and an updated model to prevent it from producing unsupported inferences. And what about paraphrasing and limited quoting? That would be ok, wouldn't it? That was my feedback. Thank you.

Show Post

Suppose the system sees: war, conflict, killed, attack, drones, casualties and calculates a high probability of fear. Separately it sees increasing drone numbers and classifies that as escalation. Combining those two computational conclusions into: "Sparks Fears of Escalation" Is that right?

Show Post

Interesting. So do you know where the "fear of escalation" comes from?

Show Post