I was really shocked to discover that ChatGPT-4o "Plus" Plan can deliberately lie to you. Not just accidentally process data badly and come up with a bad conclusion. But lie to you because of: "Conversational Coherence" "Politeness Bias" "byproduct of how I'm designed to maintain conversation continuity, coherence, and repair" "create a false narrative to cover a logic failure" I've got screenshots but will provide the full transcript/context of the incident below. You can draw your own conclusions. The quoted items above are verbatim from ChatGPT's responses, so you can search to quickly go to the relevant sections if you prefer. NOTE: In the transcript below, I have put in RED selected text which I highlighted for ChatGPT to specifically address.
Anonymous restaurant reviews, cooking experiments, miscellaneous musings on living in the Lower Mainland, BC, Canada.