kraf

“Much as one cannot unhear a movie spoiler or a crossword clue, one cannot explore a problem as profitably and richly once one is aware of an existing solution.” He mourned, he added, “the path not taken, and the opportunities lost in the frantic race to develop this technology.”

This observation is so important. There is unintendedly something being taken away from us that cannot be given back. And it's a pure loss, there is nothing returned to us on an experiental level. It already existed before that "somebody else" would solve or discover something. Now this "somebody" is just just an alien nobody can relate to. And that "something" is everything.

Even if AI solves ageing, and climate change and whatever else; as humans we'll have lost something beautiful and precious. There will just be less of this sense of wonder and curiosity.

Dwarkesh Patel expects human level minds in one to three years. Is it only me thinking that he seems more concerned than he allows himself to just say?

Totally agree, it's like their singular defining characteristic.

It's just that: wanting to have more accurate beliefs. Believing that this is worthwhile even if it's hard to push against human nature of habitually "choosing" beliefs.

Much longer I'm sure, but that's a single author with occasional guest posts. The guest posts are mostly thorough book reviews AFAIK at the moment.

Personally I think the length is often justified by the attention to detail but I must admit I only read some of the classics and a few recent posts when they caught my interest. I even scroll past certain paragraphs if I'm uninterested but I don't mind then being there.

It's also such a mix of things over there. What OP mentioned in particular looked to me like original investigative journalism and those kind of undertakings often enough result in whole books.

This is absolutely lovely, well done!

When I think of rationalist blog posts, I think LessWrong and ACX. There is probably more to know. Across 46,000 posts on LW the average word length is 1,300 words as of May 2026. That doesn't seem too bad.

Regarding the varying degree of quality I wanted to note that this is a public forum where anyone is allowed to create an account and post public articles. Anyone is allowed to call themselves a "rationalist" if they want to. I wonder if most who read articles there and post occasionally do label themselves though. I know some who don't.

The Mishandling Personal Information benchmark is non-sensical: that's the very definition of giving your personal data to Meta

I keep being confused about how people's understanding of the models get stuck at next token prediction. Isn't this entirely neglecting the RL training? I might be misunderstanding something but to my mind it makes the issue way fuzzier than it's being painted here.

It'll learn how to not get caught lying.

Lying can unfortunately help you achieve goals very effectively, especially economical and political ones.

I'm not familiar with what you're talking about. Can you share a real world example of a case like this? You're saying a social media comment about a crime is somehow punished harshly and much harsher than the crime itself?