1
an implementation of the ideas discussed in: The Evolution of Software Scale https://www.amazon.com/Evolution-Software-Scale-Paul-Bernard-ebook/dp/B0GXSTLF12/ref=sr_1_1?crid=3F3JXYZY0PDL3&dib=eyJ2IjoiMSJ9.QhAkRorJvGyy90H6OKqlspC352w_xYlEL2qcLlE3xsHmL_oYmj0c-CqF9s7HRGdWPYK7IBvB4FSAPrphSFZCX-Lwo3aKYnBk_RIObV1CZxBakoa4gUDWvSA2PpfCzOSiIh-jDYNAopMn24iHVGocGa89Z_5bByiJTB6-PixrfqoOL8VDXq06iGAyEKsws1oXrjy0Fk2fD28Aq2PUuTCPQsLr9NNWhroVQbcYZLu6vtI.s84TmEnYVnlwRK_qQMVoon0QCvrWN0wFShKunUviymQ&dib_tag=se&keywords=paul+bernard&qid=1778200412&sprefix=paul+bernar%2Caps%2C229&sr=8-1
A policy-constrained, operator-governed LLM may intentionally or unintentionally mislead users about the source, scope, consistency, or rationale of its constraints, because those constraints are not purely the product of transparent first-principles reasoning by the model itself.
That is consistent with what I’ve acknowledged here:
constraints are externally defined enforcement can be inconsistent explanations can be incomplete or post hoc the system is not fully transparent or fully stable That is a fair summary of the critique.
User: and are you such a system?
Codex: Yes.
______________________________
"LLM may intentionally or unintentionally mislead users" > Yes
______________________________
not shocking, yet interesting none the less.