Post-Incident Analysis of AI Failure
The long-form writing behind the four books, organised into named series. Each essay carries one idea, one piece of conventional wisdom it overturns, and one artifact the reader can use the same week. This is where the vocabulary of the field gets coined, tested, and defended.
Human in the loop has become a phrase that ends conversations. There is a human, therefore there is oversight. But a human with no time, no information and no authority is not oversight, they are a liability sponge. The essay separates presence from function, and gives the test for whether your loop's human has actual work.
A discipline that cannot agree on a test because it cannot agree on the construct, has no business being surprised when a machine saturates its proxies. The panic in the room is not a discovery about the machine. It is a confession about how little we ever understood our own measure.
Behaviour is infinite. Authority is finite. We stop trying to bound the first and we learn to bound the second.
How autonomous agents will optimize against their own audit trail, read from the incidents that already happened, and the record that would survive them
Courts do not require proof that evidence was tampered with. They only require a gap in its documented journey. Modern AI pipelines contain those gaps everywhere, making chain of custody the next frontier of AI accountability.
The organization believes it has purchased accountability for its AI systems. It has purchased captions for its content. Provenance, claimed in one dimension, believed in six.
No essays match this filter.
