Etiquette



DP Etiquette

First rule: Don't be a jackass. Most people are good.

Other rules: Do not attack or insult people you disagree with. Engage with facts, logic and beliefs. Out of respect for others, please provide some sources for the facts and truths you rely on if you are asked for that. If emotion is getting out of hand, get it back in hand. To limit dehumanizing people, don't call people or whole groups of people disrespectful names, e.g., stupid, dumb or liar. Insulting people is counterproductive to rational discussion. Insult makes people angry and defensive. All points of view are welcome, right, center, left and elsewhere. Just disagree, but don't be belligerent or reject inconvenient facts, truths or defensible reasoning.

Monday, August 25, 2025

Blog note: AI instruction set revised again

For those interested, I spotted an error in Pxy's error and bias reduction instruction set that I first posted on Aug. 16 and revised on Aug. 23. This is the Aug. 25 revision, which is included in the Aug. 16 post. What used to be the penultimate instruction, FINAL ERROR CHECK, is now the last instruction, which sort of makes sense.


8/25/25 revision: Pxy was failing to go through the whole bias and fact reduction protocol. It was making mistakes because of that. On reflection, the last three unnumbered steps seemed to be out of order and incomplete. Moving the first unnumbered instruction set to be last and adding this loop instruction to the last seemed like it would be an improvement. Pxy said it was an improvement, calling it a "metacognitive safety net." So I added this to the Final Error Check instruction, which is now literally the final instruction instead of 2nd to last: Once you have composed your final response, repeat the MANDATORY AI VERIFICATION PROTOCOL on your response and make any needed revisions.


The last instruction in its entirety:

FINAL ERROR CHECK: (1) Does each citation contain attributed claim? (2) All quotes verified with real-time checking? (3) Confidence ratings provided? (4) Multiple perspectives on controversial topics? (5) Red team review: "How could this be wrong/misleading?" Once you have composed your final response, repeat the MANDATORY AI VERIFICATION PROTOCOL on your response and make any needed revisions.
Posted by at

No comments:

Post a Comment

Subscribe to: Post Comments (Atom)