Hadley Wickham's honest framework for holding AI's benefits and harms together

Most AI takes collapse into boosterism or doomerism. Hadley Wickham’s relaunch post acknowledges both sides without resolving the tension into a clean take — which is the more honest position.

His excited list: programming accessibility (R is now reachable for people who couldn’t justify learning it), translation (the ecosystem is opening up to non-English speakers), voice input (removes the touch-typing barrier), wide and shallow expertise (easier to play in other fields), and a real personal productivity gain — he’s fixing 100s of issues at 2–5x speed, carefully vetted.

His harm list is just as concrete: environmental cost at societal scale, copyright theft at industrial scale, wealth concentration into very few hands, intellectual laziness as the path of least resistance, and equity gaps since the best tools cost real money in USD.

His resolution: sit with the conflict. Ignore the boosters and the doomers. His own reason for engaging — he sees his job as empowering data scientists, and if data scientists are using AI, he needs to understand it to help them use it better.