Can you make your description of how exactly your program makes it safe to run claude code on my computer more clear? When I skim read the existing description, it talked too abstractly, and also talked too lengthily about some weird dangers such as Claude convincing someone (convince whom? me? some other AI agent?) while not talking enough about actual technical sandboxing. I just want to prevent the LLM from reading any non-whitelisted files on my computer, modifying anything on my computer, and from hacking my computer. I also want to be able to let it use some USB devices or something if I need that for development. Disclaimer: I typically use openai codex instead of Claude.
philip_b
Karma: 1,013
I assume
familyhere is a function that accepts an index and returns a nonempty set? But then it seems that for the first encountered index the choice function will just return a set rather than an element from it, no? Or am I misreading something?
The idea about digging for the root cause and treating it is useful. But the idea that the root cause is always outside the body is stupid.
Correctly making a prediction involves mentally holding a whole situation in your mind at once.
Either through chunking or through a mechanism I don’t have a word for but which works exactly like chunking, you can train your personal RAM. Hold the most complex idea you can hold, then make it more complex. You will start dropping parts of the idea. Pick them up again. If you are doing it correctly, it subjectively feels like stretching. Repeat until desired mind size is achieved.
This seems similar to how I felt when I was learning to play Go by solving Tsumego puzzles. They are puzzles where you have to figure out how to kill an opponent’s group of stones or how to make your group of stones live. To do this, you need to figure out all possibly sensible moves, for each of those cheap all possibly sensible following moves, etc. I kept running out of (my brain’s) RAM. But over time I kept getting better and better at holding more complicated partial Tsumego solutions in my brain.
That is, a human plays a complete game against another human (with neither of them consulting AI during the game), then goes back to review all of their own moves (or just those flagged by AI as particularly bad), comparing them to AI’s suggestions for what those moves should’ve been, as well as looking at AI’s estimates for how bad specific human moves were.
That’s exactly the recommended (by Go players) way to review games with AI. Well, you’re also supposed to try to figure out WHY the AI-suggested moves are better than the human’s worse moves.
I have Russian citizenship. Even though I live in another country, banks really don’t like Russians having bank accounts in them and can close the account for any random reason such as a “suspicious” transaction. Also, whenever I do a SWIFT transaction between banks of different countries, the fee is typically between 20 and 100 USD, not easily predictable in advance, and whether the transfer will reach the destination or silently bounce back after hanging there for a few days is anyone’s guess.
I use cryptocurrencies to make international money transfers without any of the bullshit, risks, or fees of banks.
A comment by cousin_it that was deleted after I had made my reply.
How did you do the dopamine detoxing?
Jimi Hendrix revolutionized guitar playing.
He died from overdose at 27. Perhaps practicing for 8 hours a day wasn’t so good for his psyche?
Also, 2-10% of pro musicians get focal dystonia (a neurological condition where their muscles malfunction when they start doing their thing with their instrument), probably from playing too much. For example, the famous bassist Victor Wooten got it.
Also, I read about a famous trumpet player, maybe Louis Armstrong, and he had serious problems with his lips from playing really hard and playing a lot. He had to use a razor to cut scar tissue from his lips every day before concert to keep playing.
Also, maybe it’s possible for these weirdos (not in a bad sense) but not for a typical human, even a very smart human. At least I don’t find it possible for myself. I get tired, my muscles get tired, my ears get tired, my brain gets tired.
Also, novices at an instrument will have even more problems with unadjusted muscles and thus risk of overuse trauma, etc.
I mean for most instruments you cannot, with exceptions like the triangle. About kazoo—isn’t basically vocals with an extra sound effect? So its difficulty would be bounder from below by the difficulty of learning to sing.
Also, do you have expert level performance in any other physical skill?
Nope
I would be very surprised if you are anywhere close to 6 months of full-time deliberate practice in those 3 years?
So, a year has 255 working days. So half a year is 255⁄2 times 8 = 1020 hours of working time if we assume 8 hours per day. Yes, I probably do have approximately that amount of practice under my belt, maybe a little more. Also I’m not sure that 8 hours of deliberate practice per day for half a year in music learning is possible and is good for you.
There also a very important mental skill too. It’s not just learning to move your muscles precisely, correctly and in time. It’s also learning to hear and understand what you hear, to quickly identify what notes/chords/scales you’re hearing, what comes next, quickly generating a potentially good sounding melody and quickly understandibg of what notes it consists, what techniques need to be used to play it, where all its notes are on your instrument. Although this latter part probably cross-trains very very well between different instruments, but for a non-musician I expect it cannot be learned in 6 months.
Oh, but then you’re making a 5th category for “everything music related” :)
Ok, MAYBE someone who has already become an expert in a musical instrument and in music in general can become an expert in another musical instrument in 6 months if studying that full time, but am not at all sure about this.
So far I’ve been learning one musical instrument for 3 years, another for a year and another for a year. And I quite far from becoming an expert in any of them. Also, I doubt that you can productively practice it full time or even 4-5 hours a day because in my experience more practice per day gives diminishing returns—whether it’s because of muscles getting tired or because of brain capacity or because of boredom I don’t know.
Well, ok what I am sure of if that you have no experience in music, then becoming an expert in 6 months in any single instrument is impossible unless you’re born as a genius. This is how it is in my experience and this is what observe among musicians around me—nobody gets good quickly.
Counterexample: you cannot become an expert in playing any musical instrument in 6 months. (unless you’re a freaking genius)
I have a very mild OCD. For example, right now one of the keycaps on my laptop’s keyboard is broken, if I press it too hard, it will fly away and I’ll have to look for the metallic spring that was underneath it. So, every minute or so I feel compelled to focus my sight on that key and to touch/press it slightly to check if everything’s alright with it. Another example: when I’m walking outside alone, I regularly look at my shoes to check if my shoelaces are still tied (when I was a child, I usually tied my shoelaces incorrectly and they got untied quite often). From time to time, maybe every 10 minutes or so, I hum a simple melody, one of the few I have, under my nose. Sometimes when I sit at my laptop and my slippers are lying on the floor somewhere nearby, every few minutes I feel compelled to look at them to check if they’re still there. I think I often blink extra hard or extra long to make sure that everything is alright with my eyes, or something like that. When cooking food, I wash my hands between various actions more often than is really necessary. All of this doesn’t really bother me too much. For all these examples except the handwashing, the only downside is that if someone notices this about me, they will correctly infer that my nervous system is a bit weird. Maybe I have other mild compulsions that actually mess with my life? I don’t know. I don’t really have a sense of panic when any of this happens or when I ignore the urge. Just a slight discomfort. So, any advice for me?
I also would like to discuss your methods of dealing with OCD. First, let me note that I predict that your current methods won’t work well forever and that problems will return, maybe transformed into slightly different problems. So, anyway, about the example with eye contact with beautiful women. It is not in fact the case that any amount of eye contact with women in any situation is optimal. I think in some cases it is actually significantly suboptimal—sometimes it can scare them, or weird them out, or send them or someone else a wrong signal, etc. I think for a healthy person, they would just learn intuitively when and how to do it and when not to do it. But, as I understood, your OCD-managing methods basically allow any amound of eye contact in any situation. That seems really really crude. As I understood, you basically decided “ok, my emotions and instinctive reactions regarding eye contact with beautiful women don’t work at all, so it’s better to ignore them completely”. Well, I think that’s very suboptimal compared to if you had healthy emotions and instinctive reactions and listened to them. Because sometimes in some situations, the instinctive desire to avert the gaze is actually right. So, what are your thoughts and your plans regarding this? And in general, you label thoughts as “worrying” or “threat monitoring”, etc. and then ignore them. Ok, for clear cut situations like beeping, where on a rational level it’s obvious that there is actually no threat, that seems good. But what about less clear cut situations? I mean, I except that at some point you’ll have dealt with all such low hanging fruit with really super obviously suboptimal behaviour. But then you’ll probably want to deal with less obviously suboptimal behaviour, where the sense of anxiety or panic or whatever might actually be right or at least might be pointing in the right direction. What then?
Ok, as a non-american and non-chinese, I notice that these last few years very very often Americans say or imply that China is aggressive towards other countries, that China is at a war through technologies with the US and that it will be very bad for the US if China builds advanced AI first. Can anyone explain to me why this is the case?
Ok, I’ve looked through ~75% of emails mentioning “beef jerky” on https://jmail.world/search?q=beef+jerky. Haven’t seen anything weird in them, no euphemisms, nothing. They just talk about beef jerky cooked by Francis Derby, who used to work as a chef in a restaurant called Cannibal.
That’s exactly how I woke up today. For me, a normal kind of shower (a long one) and changing my clothes works, kinda.
Can you explain that? What’s the difference between those two things?