Preprint
Aug 2026
Understanding as an Explicit and Assessable Component of Frontier AI Safety Decisions
The provisional methodology for making understanding explicit and assessable requires the production of an explicit description of 4 objects of understanding and a justification for the adequacy of this understanding and it is found the analyses that justify sufficiency of understanding drives the engineering.
Stephen Barrett, Robin Bloomfield, A. Chirilă et al.
· 0 citations