AI just repeats whatever the prevailing opinion is at that time. I am a very heavy AI user (Claude, Gemini, ChatGPT) and have queried it on a variety of topics. AI is not thinking, it is repeating.
But that is not what most “work” usually means. Work is mostly repetitive. The actual moment of decision is brief.
So what do I mean by work here? I mean the collection, organization, and synthesis of the materials needed before reaching that decision.
For that part of the process, AI is extremely effective.
Otherwise they would
1. force humans to spend much more time on these decisions, and
2. Teach humans that the only way to save time on decision-making is to collaborate with other humans
I had a reply to your Asian take https://news.ycombinator.com/item?id=47970518
That would be an improvement. They are generally far too sychophantic to just repeat the prevailing opinion, and instead synthesise the opinion that they think wants to be heard by the user.
This^. We aren’t talking about general intelligence. We are mostly using a large language model. It’s a model of languages and not the universe. Language encodes a lot, enough to be “unreasonably effective”.
I'm progressive, but this is just not helpful. Progressives should focus less on trying to argue that AI sucks actually (it obviously doesn't), we should focus on thinking about how to transform society to still protect human quality of life and dignity even when strong AI exists.