PJ2
Gets better. A few months back, another expat captain was riding as P2 (of a double crewed Ultra LH flight) on the jump seat for TO, on the daily Air India BOM-JFK (16+ hours flight). TOGA was selected to begin the TO run on R/W 27 but Auto Throttle (A/T) did not engage so instead of setting thrust manually, the commander as PF selected A/T on the MCP as he rolled down the runway. Only thing was, he inadvertently selected Autopilot, which duly engaged (the industry was going through a spate of these events at the time) and as both he and the 250 hour ace in the RH seat failed to check their FMAs, the Max AUW aircraft, with 112 tons of gas on board, now thundered down the runway with no means of rotating and hence getting safely airborne.
Expat captain had noticed the above events as they unfolded (easily done from the jump seat) and had started out conversationally pointing the situation out. On getting no response, his cries became louder (and shriller) to the effect that Autopilot should be disengaged until in the end, in mortal peril, he found himself shouting at the top of his voice to disengage the Autopilot and un-strapping to reach over and do the damned thing himself. Commander then seemed to wake up, hastily disengaged the Autopilot and snatch rotated near the end of the runway, narrowly avoiding smashing through the boundary fence as a flaming fireball, carving across the traffic choked main road and exploding amidst the dense residential housing beyond.
During the subsequent uneventful cruise, an extremely shaken expat captain eventually pointed out that an Incident Report had to be filed so that the airline and the industry could learn the lessons available but the commander remained evasive and non-committal, presumably not wishing to go through an upcoming Safety Department session of “identify culprit, rigorously interview, belittle, apportion full blame, excoriate, punish”. In the event no report was made but the expat captain reportedly made a confidential FAA/NASA submission back in the US.
I think that we can all draw our own conclusions on the current AI Safety philosophy.