AI is just automation

This talk by Stuart Halloway makes a great point: AI is just automation; it carries the same disruptive potential, both positive and negative, as previous automations before it.

He asks the questions that are valuable to software engineer beyond the reflex “does it work?” – and give convincing answers:

• What is AI? -> just a new form of automation. No existential risk, no conscience.
• Why is it (AI hype/using token output as a software quality metric) happening? -> because management thinks they can get away with it, the same way they got away with imposing such unrealistic objective to Amazon workers that they had to pee in bottles
• Who does it benefit? -> the ultra rich, with the help of lawmakers (cf. 2025 inauguration)

For all lazy (and therefore good) programmers, automating as much as possible with AI is enticing. But it comes with a significant loss of skill in autonomy (cf. the study on schoolchildren’s drop in tests results after allowing GenAI for homework). Cognitive surrender has to be resisted actively: deciding when to take back control from coding agents and enforcing it, to gain the practical knowledge and experience that comes with friction.