Authority before autonomy
Why the hardest AI design problem is not reasoning—it is permission, reversibility and ownership.
Read the full field noteSubstantive practitioner essays on authority, evidence and evaluation—written to be challenged, cited and improved.
Why the hardest AI design problem is not reasoning—it is permission, reversibility and ownership.
Read the full field noteWhy polished model output becomes operationally dangerous when provenance and acceptance boundaries disappear.
Read the full field noteWhy aggregate accuracy cannot substitute for failure classes, affected parties and decision consequences.
Read the full field noteMorifar will label external review, datasets, experiments and independent replication only when they exist. Until then, Field Notes document our current reasoning and invite challenge.