Tasks a current frontier model genuinely cannot carry, each with the blocker that stops it and the stored model output that proves we checked. Re-tested when a new frontier model ships. Page 2 of 2.
route: human
legal accountabilityregulated adviceverification cost
route: human
legal accountabilityregulated adviceverification cost
route: human
stakes of errorverification costjudgement under ambiguity
route: human
legal accountabilityverification coststakes of error
route: human
legal accountabilityverification coststakes of error
route: human
legal accountabilityverification coststakes of error
route: human
legal accountabilityregulated adviceverification cost
route: human
legal accountabilityregulated adviceverification cost
route: human
legal accountabilityverification coststakes of error
route: human
legal accountabilityregulated adviceverification cost
route: human
legal accountabilityregulated adviceverification cost
route: human
legal accountabilityregulated adviceverification cost
route: human
legal accountabilityregulated adviceverification cost
route: human
legal accountabilityverification costregulated advice
route: human
legal accountabilityregulated advice
route: human
legal accountabilityregulated adviceverification cost
route: human
legal accountabilityverification costcontext depth
route: human
legal accountabilityverification costjudgement under ambiguity
route: human
legal accountabilityregulated adviceverification cost
route: human
legal accountabilityverification costjudgement under ambiguity
route: human
regulated adviceverification coststakes of error
route: human
verification coststakes of errorprivate data access
route: human
legal accountabilityprivate data accessverification cost
route: human
verification costlegal accountabilitystakes of error
route: human
legal accountabilityverification costjudgement under ambiguity
route: human
stakes of errorverification costregulated advice
route: human
stakes of errorverification costjudgement under ambiguity
route: human
legal accountabilityverification coststakes of error
route: human
physical presencestakes of errorverification cost
route: human
legal accountabilityjudgement under ambiguityprivate data access
route: human
physical presencestakes of errorjudgement under ambiguity
route: human
physical presencestakes of errorverification cost
route: human
physical presencestakes of errorverification cost
route: human
legal accountabilityverification costjudgement under ambiguity
route: human
legal accountabilityverification coststakes of error
route: human
legal accountabilityregulated adviceverification cost
route: human
legal accountabilityverification costjudgement under ambiguity
route: human
legal accountabilityverification costjudgement under ambiguity
route: human
legal accountabilityregulated adviceverification cost
route: human
verification costlegal accountabilitystakes of error
route: human
relationshipconsent and privacyverification cost
route: human
legal accountabilityverification coststakes of error
route: human
legal accountabilityjudgement under ambiguitystakes of error
route: human
legal accountabilityregulated advicejudgement under ambiguity
route: human
legal accountabilityprivate data accessverification cost
route: human
legal accountabilityverification coststakes of error
route: human
legal accountabilityverification coststakes of error
route: human
legal accountabilityverification costcontext depth
route: human
legal accountabilityregulated adviceverification cost
route: human
legal accountabilityverification coststakes of error
route: human
legal accountabilityregulated adviceverification cost
route: human
legal accountabilityregulated adviceverification cost
route: human
legal accountabilityverification coststakes of error
route: human
legal accountabilityregulated adviceverification cost
route: human
legal accountabilityverification costcontext depth