The overall essay leans toward telling users to fight technology and stay "old-school" reliable.
I think his aviation argument fights that premise though.
@AI is flying a modern Dreamliner safer than flying a 747 from 1970?
Yes, by a wide margin, though the comparison isn't clean. Fatal accidents per million flights fell from about 6 in the 1970s to about 0.5 in 2019, roughly a twelvefold drop (Aviation Safety Network figures, as summarized on Wikipedia's aviation safety page). The IATA's five-year rate for 2021 to 2025 is one fatal accident per 5.6 million flights, and its 2025 all-accident rate was 1.32 per million flights (IATA 2025 Safety Report, published March 2026: iata.org/en/pressroom/2026-releases/2026-03-09-01/).
For the aircraft themselves, Airbus's accident statistics split fleets by generation. Its fourth generation, which includes the 787, had a fatal accident rate of 0.04 per million flight cycles in 2025, against 0.09 across all generations that year. That's a fair comparison of airframes, but not of a 1970 747 against a 787 alone, since the 1970 747 was a first-generation jet and the figure also reflects airline training, air traffic control, weather forecasting and runway standards.
One caveat on your question: a Dreamliner isn't flown by AI. Its autopilot and flight management software are deterministic rule-based automation, which is the kind of automation the essay is about. The safety gain comes mostly from better engines, materials, redundancy and crew procedures, and the rare failures the author worries about are still the ones that matter.
