All of npostavs's Comments + Replies
What about tuning the fiddle strings down 1 tone?
You say this:
If you’re thinking, “Wait no, I’m pretty sure my group is fundamentally about X, which is fundamentally good,” then you’re probably still in Red or Blue.
But you also say this:
First, the Grey tribe is about something, [...] things that people already think are good in themselves.
Doesn't the first statement completely undermine the second one?
I guess you meant jukebox, not jutebox. Unless there is some kind of record-playing box made of jute fiber that I haven't heard of...
but I recently tried again to see if it could learn at runtime not to lose in the same way multiple times. It couldn't. I was able to play the same strategy over and over again in the same chat history and win every time.
I wonder if having the losses in the chat history would instead be training/reinforcing it to lose every time.
Yes, my understanding is that the system prompt isn't really priviledged in any way by the LLM itself, just in the scaffolding around it.
But regardless, this sounds to me less like maintaining or forming a sense of purpose, and more like retrieving information from the context window.
That is, if the LLM has previously seen (through system prompt or first instruction or whatever) "your purpose is to assist the user", and later sees "what is your purpose?" an answer saying "my purpose is to assist the user" doesn't seem like evidence of purposefulness. Same if you run the exercise with "flurbles are purple", and later "what color are flurbles?" with the answer "purple".
#2: Purposefulness. The Big 3 LLMs typically maintain or can at least form a sense of purpose or intention throughout a conversation with you, such as to assist you.
Isn't this just because the system prompt is always saying something along the lines of "your purpose is to assist the user"?
1by saying their name aloud: [...] …but it’s a lot more difficult to use active recall to remember people’s names.
I'm confused, isn't saying their name in a sentence an example of active recall?
Finding two bugs in a large codebase doesn't seem especially suspicious to me.
I don't think I understand, what is the strawman?
I think the AI gave the expected answer here, that is, it agreed with and expanded on the opinions given in the prompt. I wouldn't say it's great or dumb, it's just something to be aware of when reading AI output.
It looks like you are measuring smartness by how much it agrees with your opinions? I guess you will find that Claude is not only smarter than LessWrong, but it's also smarter any human alive (except yourself) by this measure.
Entries 1a and 1b are obviously not not relevant to the OP, which is mainly about the sense in 3b (maybe a little bit the 3a sense too, since it is "merged with or coloured by sense 3b").
Entry 3b looks (to me) sufficiently broad and vague that it doesn't really rule anything out. Do you think it contradicts anything that's in the OP?
The OED defines ‘gender’, excluding obsolete meanings, as follows:
Okay? Why are you telling us this?
Maybe if you solve for equilibrium you get that after releasing the tool, the tool is defeated reasonably quickly?
I believe it's already known that running the text through another (possibly smaller and cheaper) LLM to reword it can remove the watermarking. So for catching cheaters it's only a tiny bit stronger than searching for "as a large language model" in the text.
Why release a phone with 5 new features when you can just R&D one and put it in a new case?
In the ideal case of a competitive market, you don't release just one new feature, because any of your competitors could release a phone with two new features and eat your lunch. But the real-world smartphone market is surely much closer to oligopoly than perfect competition.
The costs of the competition of the market are almost invisible, but we have been seeing them over decades get more and more obvious.
How sure are you that this isn't rather the costs of lack of competition?
Maybe, although what is "sufficient" depends a lot on the rate of catching the evaders. I don't have a good guess as to what that rate is.
Yes, currently very few companies report paying ransom payments. When this tax is introduced the motivation for hiding payments will be even higher, and go up with the tax rate. So when you say "With each increase in tax rates, a market equilibrium will be reached where the funding of ransomware is significantly reduced" I would guess instead that reporting will go down.
You didn't say anything about tax evasion in this post, which seems like an important thing to consider. Most ransomware payments are made secretly, right?
Worsening housing and rent problems in California, Canada, major metropolitan areas, Japan, China, and other places that are facing housing shortages could ignite support for Georgism.
Do Japan and China have housing shortages? I thought Japan was the canonical "zoning done right" example. And doesn't China have some sort of over-supply sitatution due to government subsidies?
slatestarcodex being contra hanson on healthcare
That case (I didn't follow the others) seemed like it was mostly about confusion over what Hanson's position even is. Maybe because Hanson and/or people misunderstanding him tried to compress it into short tweets.
But how can you know that? Couldn't there be actual insider sources truthfully reporting the existence of such discussions?
Yes, I perhaps should have said "I think there is a 99% chance this is made up". As a general rule, I think any politically charged story based on "anonymous insider sources" should be considered very low credibility, and if there is no other support, then a 90+ chance of being made up is about right. More credibility points lost in this case for the only source being a tweet from a guy who seems to be advertising some kind of pass...
Canada also is looking to impose a $25k penalty and double its ‘exit fee’ for citizens who leave the country, to ‘curb the emigration crisis.’
This is made up, apparently.
https://thezvi.substack.com/p/monthly-roundup-18-may-2024/comment/56269684
https://www.yahoo.com/news/users-spread-unfounded-claims-impending-163724801.html
Recent headlines are about too much immigration (e.g., https://www.theglobeandmail.com/business/article-canada-stuck-in-population-trap-needs-to-reduce-immigration-bank/), so 'emigration crisis' doesn't make much sense.
Unless you also think the United States is an outlier in terms of spouses who don't unconditionally love each other, I guess you have to endorse something like Kaj_Sotala's point that divorce isn't always the same as ending love though, right?
probably the majority of spouses unconditionally love their partners.
How do you square this with ~50% of marriages ending in divorce?
a good trade for immunity to cavities and gum disease.
If you throw in immunity to bad breath
FYI, https://www.luminaprobiotic.com/faq says used to say
This strain doesn't do anything to protect against gum disease, or bad breath.
And he thinks Hermes 2 Pro is ‘cracked for agentic function calling,’
I don't understand what the word 'cracked' means here; "broken" or "super awesome" or ...?
persuade/inspire/motivate/stimulate etc is just the politically correct way of saying what it actual is, which is manipulation.
Persuade has a fairly neutral connotation for me, that is "I was persuaded to give 10k to a scammer" and "I was persuaded by a friend to quit my day job" both seem correct to me. I would nominate that as the word for describing what it "actually" is, rather than "manipulation" which seems overly negative/cynical.
I think anorexia is in a different category because the patient often doesn't want to get better. David Burns talks about it a little on https://feelinggood.com/2019/11/25/168-ask-david-the-blushing-cure-how-to-heal-a-broken-heart-treating-anorexia-and-more/, where he mentions that some sort of therapy with a 50% success rate is good.
The rapid cure stuff is mainly about depression and anxiety disorders, I guess agoraphobia should count (with the caveat that the patient has to be well enough to reach the therapist's office). Certainly whether it "could take...
1David Burns also has his own podcast, many episodes of which are example live sessions of this rapid cure (see https://feelinggood.com/list-of-feeling-good-podcasts/ and search for "live therapy", or https://feelinggood.com/podcast-database/ which has a fancy Javascript interface allowing filtering on tags).
He does often make the explicit claim on his podcast, that 90% of patients can be cured in one or two sessions (plus one more for "relapse prevention"). It's a bit hard to know how much of this is from a selection effect on the patients though. I'm pret...
A big part of understanding the culture of futility is understanding how traumatic it is when the bad guys win. When SBF, the Luke Skywalker of crypto, and CZ, the Darth Vader of crypto, go head to head and CZ emerges victorious. Then CZ says "Ha! serves you right for being an idiotic do-gooder" and everyone cheers.
Didn't we actually learn that they were both bad guys? I find this example confusing.
I was kind of surprised by this too; I found this study which seems to support it though: https://theconversation.com/we-studied-what-happens-when-guys-add-their-cats-to-their-dating-app-profiles-144999
...In our study, we recruited 1,388 heterosexual American women from 18 to 24 years old to take a short anonymous online survey[...] Most of the women found the men holding cats to be less dateable. This result surprised us, since previous studies had shown that women found men with pets to have higher potential as partners. They also thought the men holding
The NYT paywall doesn't didn't do anything if Javascript is disabled.
EDIT: I've noticed recently that NYT articles are cut-off before the end now, even without JavaScript. I wonder if the timing of this paywall upgrade is related to the lawsuit?
No particular reason why we can only have 42 chromosomes
Isn't having extra chromosomes usually bad? https://en.wikipedia.org/wiki/Trisomy
(PS the usual number is 46)
What is an example where two negative numbers multiply to give a negative number?
Since you didn't specify real numbers, it seems like -i * -i = -1 should fit?
We know roughly how to achieve immortality
Isn't the assumption that once we successfully align AGI, it can do the work on immortality? So "we" don't need to know how beyond that.
then you could spread the pesticide (and not other pesticides) in the region
This would affect other insects in addition to the targeted mosquitoes, right? This seems strictly worse than the original gene drive proposition to me.
A survey shows that gay male teenagers are several times more likely to conceive girls than straight male teenagers.
Does "conceive" mean "have sex with" here? Because according to what I think of as the standard definition of that word, you would be saying that gay male teenagers are more likely to produce female offspring (which sounds pretty silly). Did the survey use that word?
Also asked (with some responses from the authors of the paper) here: https://www.lesswrong.com/posts/khFC2a4pLPvGtXAGG/how-to-catch-an-ai-liar-lie-detection-in-black-box-llms-by?commentId=v3J5ZdYwz97Rcz9HJ
Testing with PortAudio's demo paex_read_write_wire.c [2]
It looks like this uses the blocking IO interface, I guess that adds its own buffering on top of everything else. For minimal latency you want the callback interface. Try adapting test/patest_wire.c or test/pa_minlat.c.
Humans have lived during one of Earth's colder period, but historically it's been a lot hotter. Our bodies are well adapted for heat (so long as we can cool off using sweat)
This doesn't seem very reassuring? For example, https://climate.nasa.gov/explore/ask-nasa-climate/3151/too-hot-to-handle-how-climate-change-may-make-some-places-too-hot-to-live/
...Since 2005, wet-bulb temperature values above 95 degrees Fahrenheit [35 C] have occurred for short periods of time on nine separate occasions in a few subtropical places like Pakistan and the Persian Gulf. T
Let me just quote Wikipedia: "A seahorse [...] is any of 46 species of small marine fish in the genus Hippocampus." Because I spent a few confused minutes trying to figure out how males could face more intense competion in a brain part.
2He says non-programmers; I guess you misread?
Theoretically capitalism should be fixing these examples automatically
Huh? Why?
If you want to get a job working on machine learning research, the claim here is that the best way to do that is to replicate a bunch of papers. Daniel Ziegler (yes, a Stanford ML PhD dropout, and yes that was likely doing a lot of work here) spent 6 weeks replicating papers and then got a research engineer job at OpenAI.
Wait, a research job at OpenAI? That’s worse. You do know why that’s worse, right?
I don't know why, and I'm confused about what this sentence is saying. Worse than what?
I don't think anyone is proposing to offer this deal to Putin; it's not like the rank and file soldiers are able to make the "invade your neighbor" decision in a bid to get EU citizenship.
Low confidence generally means questionable or implausible information was used, the information is too fragmented or poorly corroborated to make solid analytic inferences, or significant concerns or problems with sources existed.1
In what sense is second-order logic "beyond the reach of machines"? Is it non-deterministic? Or what are you trying to say here? (Maybe some examples would help)