- https://archive.is/uIC4O#selection-533.0-538.0
- by Ezra Klein on NYT
8/26/25
Ezra is putting aside all the societal issues associated with AI and looking at the tech itself.
He’s talking about how ChatGPT 5 is a pretty credible personal assistant.
He assumes the tech will continue to advance.
He cites two different perspectives on AI future:
- AI as Normal Technology - interacting with the external world and society will always be a speed limit on what AI can accomplish. It is compared to electricity. It took decades for the impact of the tech to start being reflected in society.
- AI 2027 - this is the more classic “AGI Super Intelligence rules the world” perspective
Both perspectives have interesting critiques of the other:
- AI as Normal Technology says those who fear AI conflate intelligence with power. We assume because we are smarter than ferrets we control the world. If AI is smarter than us it will control the world. In reality, intelligence is not a 1:1 translation to power. It took humans eons to build the technological civilization we have today.
- AI 2027 perspective is that “normal technology” paper talks about how hard it is for tech to propagate through bureaucracies. We still cannot always share medical info between doctors easily. AI tools have spread very very quickly in adoption, because it is so general it is easy to slot into many workflows.