LessWrong 2.0 Reader

It's also available on Android.

FWIW I think it probably would be, between those two. Land and houses are different, even if we usually buy them together. When I bought my first house, the appraisal included separate line items for the house vs the land it was on, and the land was a majority of the price I was paying. I don't know what the OP actually meant, but to my own thinking, owning land (in the limit of advanced technology making everything buildable and extractable) means owning some fixed share of Earth's total supply of energy, water, air, and minerals. Building a house, given the space and materials, might become arbitrarily cheap in the future through automation, but competition for space and materials might become arbitrarily intense as the number of competing uses for them increases. Depends a lot on what order things happen in, and whether property rights remain a meaningful concept at all.

Right now, for a few hundred thousand dollars, you can build a home and (if you have enough land) buy enough equipment to make it self-sustaining in food, water, and energy (until things break). In the limit of high technology you can extend that to almost all needs, make the overall system indefinitely self-repairing, and increase carrying capacity of any given plot of land. You can also earn money by selling a flow of excess energy, which may be valuable even to an AI provided the AI respects your rights to same in the first place.

Cool, thanks!

(B’) “…Gee, I guess this reprocessing must have been a kind of ‘training / practice / exercise’ during which I could forge new better subconscious habits and associations related to ‘the feeling of anxiety’ in general. And these new subconscious habits and associations are now serving me well in a wide variety of adult contexts.”

I think that there are definitely techniques that work on one's reaction to the feeling of anxiety in general, but the specific ones that I had in mind don't feel like they'd be doing that. Rather they seem much more localized, in that they eliminate some particular anxiety trigger or specific kind of anxiety from getting triggered in the first place. But then if something else happens to trigger the same or a similar anxiety, the person isn't necessarily any better at dealing with that.

So if someone feels the same kind of anxious around both spiders and snakes, then this kind of an intervention might eliminate the fear of spiders entirely, while leaving the reaction to snakes entirely unaffected (or vice versa).

Also, AFAICT, people achieve great therapeutic success by methods that involve bringing up childhood memories, but other people [LW · GW] also achieve great therapeutic success by methods that don’t.  :)

Oh yeah definitely, didn't mean to imply that working with memories would be the only approach that worked.

My experience so far with writing all my notes in public [LW(p) · GW(p)]. 

For the past ~2 weeks I've been writing LessWrong shortform comments every day instead of writing on private notes. Minimally the notes just capture an interesting question / observation, but often I explore the question / observation further and relate it to other things. On good days I have multiple such notes, or especially high-quality notes. 

I think this experience has been hugely positive, as it makes my thoughts more transparent and easier to share with others for feedback. The upvotes on each note gives a (noisy / biased but still useful view) into what other people find interesting / relevant. It also just incentivises me to write more, and thus have better insight into how my thinking evolves over time. Finally it makes writing high-effort notes much easier since I can bootstrap off stuff I've done previously. 

I'm planning to mostly keep doing this, and might expand this practice by distilling my notes regularly (weekly? monthly?) into top-level posts

Relevant to this discussion: https://novasky-ai.github.io/posts/sky-t1/

Open source model trained with $450 of compute.

I agree that I have no faith in current governments to implement and enforce policies that are more complex than things on the order of governance compute and chip export controls.

I think the conclusion this points towards is that we need new forms of governance. Not to replace existing governments, but to complement them. Voluntary mutual inspection contracts with privacy-respecting technology using AI inspectors. Something of that sort.

Here's some recent evidence of compute thresholds not being reliable: https://novasky-ai.github.io/posts/sky-t1/

Here's another self-link to some of my thoughts on this: https://www.lesswrong.com/posts/tdrK7r4QA3ifbt2Ty/is-ai-alignment-enough?commentId=An6L68WETg3zCQrHT [LW(p) · GW(p)]

It's not just a question of whether people agree but whether they actually comply with it. People agree to all sorts of things but then do something else. 

Thanks so much for writing this! You expressed the things I was concerned about with much more eloquence than I was able to even think about.

I've come to believe that the only 'correct' way of using AGI is to turn it into a sort of "anti-AGI AGI" that prevents any other AGI from being turned on and otherwise doesn't interact with humanity in any way: https://www.lesswrong.com/posts/eqSHtF3eHLBbZa3fR/cast-it-into-the-fire-destroy-it [LW · GW]

We plan to iterate on this research note in the upcoming weeks. Feedback is welcome!

Ideas I want to explore:

• New reasoning models may be released (e.g. deepseek-r1 API, some other open source ones). Can we reproduce results?
• Do these ITC models articulate reasoning behind e.g social biases  / medical advice?
• Try to plant backdoor. Do these models articulate the backdoor?

thanks! heres my initial thought about introspection and how to improve on the setup there:

in my introspection paper we train models to predict their behavior in a single forward pass without CoT.

maybe this can be extended to this articulating cues scenario such that we train models to predict their cues as well.

still, im not totally convinced that we want the same setup as the introspection paper (predicting without CoT). it seems like an unnecessary restraint to force this kind thinking about the effect of a cue in a single forward pass. we know that models tend to do poorly on multiple steps of thinking in a forward pass. so why handicap ourselves?

my current thinking is that it is more effective for models to generate hypotheses explicitly and then reason about what effects their reasoning afterwards. maybe we can train models to be more calibrated about what hypotheses to generate when they carry out their CoT. seems ok.