AI in Safety
5 pieces, each argued from primary sources and each handing you something you can use on Monday.
-
The ScoutThe AI That Writes Your Incident Report Learned Root Cause From the Reports Before It
AI tools that draft post-incident investigation reports are pattern-matched against a corpus of past reports that already over-attributes causation to operator error, so faster drafting will scale that bias, not correct it.
-
The ScoutPredictive maintenance is optimising for uptime. Your safety layer is not.
Condition monitoring finds degradation that announces itself. A proof test finds dangerous failures that do not, which is why PdM coverage cannot be traded for a longer proof test interval without changing the PFD you certified.
-
The ScoutAn AI can draft your JHA. It cannot do it.
A language model writes a job hazard analysis from generic patterns, not from your floor, so it will omit the site-specific hazard a walkdown would catch and dress the omission in confident, standard-sounding prose.
-
The SkepticPredictive safety and the base-rate trap
The rarer the incident, the more a 'predictive' model gets wrong every time it fires, and it's trained on the injuries you already have, not the ones that kill.
-
The ScoutAn AR overlay is only as honest as the procedure behind it
Augmented reality can enforce a correct isolation sequence and cut skipped steps, but an AR overlay built on an outdated energy-control procedure is more dangerous than paper because it turns a wrong instruction into a confident one.