The AI-collapse pre-mortem
The author reflects on their April 2023 article predicting AI disruption, noting they failed to make specific predictions but correctly anticipated the impact while underestimating the scale of subsequent developments.
Related stories
The article discusses software engineering folklore, noting that safety is often determined retrospectively rather than during development. It examines how systems are labeled as safe only after they have proven reliable over time.
The article explores the concept of message-passing as a fundamental approach to data processing and communication. It examines how taking this paradigm seriously could reshape our understanding of data structures and computational models.
The article references a 1997 talk, highlighting how architectural considerations become increasingly important as systems grow in complexity, often outweighing material choices.
The author discusses how when a fundamental input becomes inexpensive, it leads to unpredictable explosions in new categories of activity. This builds on previous writing about the transformative effects of cost reductions in key inputs.
The article discusses collective superstitions among people who interact with machines, examining patterns in how humans anthropomorphize technology and develop ritualistic behaviors when communicating with artificial systems.