B) We will also have the card game Predictably Irrational. Feel free to bring your favorite games or distractions.
C) We usually walk and talk for about an hour after the meeting starts. There are two easy-access mini-malls nearby with hot takeout food available. Search for Gelson's or Pavilions in the zipcode 92660.
D) Share a surprise! Tell the group about something that happened that was unexpected or changed how you look at the universe.
E) Make a prediction and give a probability and end condition.
F) Contribute ideas to the group's future direction: topics, types of meetings, activities, etc.
Conversation Starter Readings:
These readings are optional, but if you do them, think about what you find interesting, surprising, useful, questionable, vexing, or exciting.
1)Georgism... In Space! Just and proper political economy for an interplanetary civilization
Georgism... In Space! - by Sam Harsimony
How did Georgism define economic land, and why is it important for tax policy in space?
Can you explain how Georgism can be applied to space resources like energy, matter, and physical space?
How might governments in space subsidize the collection of solar energy?
What are some difficulties with taxing matter in space, and how might taxes need to be adjusted to avoid distortions?
How can physical space in space be taxed, and what factors might be considered in assessing the value of land in space?
How can excludable resources like broadband spectrum and orbits be properly managed in space?
Why is it important to strike a careful balance when implementing taxation policies for space colonization?
I've seen talk about ending wildlife suffering, a goal which seems impractical, but while we're at ot. why not look at optimizing for pleasure for wild animals?
I'm extremely worried about AI risk lately, especially after the twin shocks of ChatGPT and Sydney. I want to do something to help. I live near DC and know people with political connections who could potentially help me get a policy meeting with a member of Congress. Two questions:
1. If I can get such a meeting, what specific policies should I propose? My own ideal is to ban all AI research, but I know that's an impossible ask.
2. Any insights on members of Congress who are particularly good targets to try and meet with specifically? I'm talking to my connections about this too, but I'm wondering if anyone here has knowledge specific to the AI risk field, in terms of which members may be receptive or have been receptive in the past, and have the power/motivation to try to do something about it.
Fucking hell, I love that G. E. Moore was dumb enough to argue that holding up his hands in front of him and saying 'Here is one hand, here is another' was enough to prove the existence of an external reality, Jesus, I am going to murder every philosophical position I disagree with with that one.
I'll start: One of my personal favourites is Norm Macdonald's gag about Bill Cosby on Comedians in Cars Getting Coffee. Everything about that joke is perfect.
I was reading through an old bookmark and saw that Scott was still looking to review Nixonland eventually. I read the book myself after that banger book review, but found that it's actually part of a quartet:
I've not read Before the Storm or Reaganland yet, but Nixonland flows near seamlessly into Invisible Bridge and I imagine the other two books segway cleanly too. I'd napkin that the whole thing is close to 150 hours of audiobook.
Even though Perlstein is quite partisan (he wrote an '05 book on how the DNC could take center stage again), I'm not sure anyone else had even attempted to write a consolidated history like this of American society/politics/discourse/mood for such periods of time.
I'd grown up around adults who made mention of things like Woodstock and Watergate or the '68 DNC riot, and even someone who still couldn't believe we'd elected an actor (Reagan) as president. It was all ancient history, with the current era being written on TV as and after the towers fell. Reading another generation's political experience connected *a lot* of dots in my zoomer mind. I'd defo push these books at any zoomer interested in politics frfr (or anyone seeking to understand the boomers).
I know there has been back and forth on this and it is not the wonder drug it was thought to be for a while but I've been taking a single aspirin before going out to shovel a large amount of snow for a couple years as a sort of voodoo ritual to protect myself.
I'm 70 now but have no diagnosed cardio issues, and have good cholesterol numbers, good blood sugar, healthy weight, good blood pressure and an echo cardiogram shows no buildup of bad stuff in my heart. In the summer I still ride decent mileage on my bike, though my 'century' rides are all metric now. I also sit and crank my old Schwinn on a turbo trainer in the winter until I think I'm risking dying of boredom.
But... when my dad was my age he died from his third heart attack so I figure taking an aspirin before I go out and perform what used to be a pretty reliable widow maker chore, hell, it can't hurt.
Does this make sense to anyone who actually knows about this stuff?
The Biden Administration is obviously going to lose in the Supreme Court regarding the student-loan cancellation. And on the merits they should lose, though this Court would rule against them regardless.
Both the president and his staff know that's what's going to happen, and privately don't actually mind it. The whole thing is kabuki theater.
"“We know that early-life stress impacts the brain, but until now, we didn’t know how,” Baram said. “Our team focused on identifying potentially stress-sensitive brain pathways. We discovered a new pathway within the reward circuit that expresses a molecule called corticotropin-releasing hormone that controls our responses to stress. We found that adverse experiences cause this brain pathway to be overactive.”
“These changes to the pathway disrupt reward behaviors, reducing pleasure and motivation for fun, food and sex cues in mice,” she said. “In humans, such behavioral changes, called ‘anhedonia,’ are associated with emotional disorders. Importantly, we discovered that when we silence this pathway using modern technology, we restore the brain’s normal reward behaviors.”
So there's this cluster of positions called "Illusionism" which is about doubting the existence of consciousness to various degrees, whatever that means exactly. I'm very interested in understanding how people think about this in more detail, so if anyone here is sympathetic to that set of ideas, I'd like to hear from you! Like, what is consciousness, what exactly does your position say about it, and why do you think that? (And if it's applicable, what do you see as the objections to that position, and why are they unconvincing?)
If it's relevant, I'm mostly through Dennett's "consciousness explained", and I think I understand his model quite well.
Why is this being done? If Dahl's works are so flawed, and so many of the passages need to be edited to the point of losing Dahl's characteristic nastiness and not even being recognizably Dahl any more, why not just toss the whole thing out? What's the point of keeping something so flawed?
The obvious answer is that modern corporations and woke writers are so bereft of genuine creative talent that even a dreadfully unprogressive straight white Englishman born over 100 years ago was creating categorically better art than all these modern 'enlightened' fools could ever dream of making themselves (or at least, if they don't recognize Dahl's actual greatness, they certainly acknowledge the enduring popularity his works have that their own works do not).
The discussion surrounding large language models (LLMs) and their relationship to AGI has been utterly horrendous. I believe LLMs and their intellectual descendants will be as transformative to society as the transistor. This technology deserves careful analysis and argument, not dismissive sneers. This is my attempt at starting such a discussion.
To start off, I will respond to a very common dismissive criticism and show why it fails.
>It's just matrix multiplication; it's just predicting the next token
These reductive descriptions do not fully describe or characterize the space of behavior of these models, and so such descriptions cannot be used to dismiss the presence of high-level properties such as understanding or sentience.
It is a common fallacy to deduce the absence of high-level properties from a reductive view of a system's behavior. Being "inside" the system gives people far too much confidence that they know exactly what's going on. But low level knowledge of a system without sufficient holistic knowledge leads to bad intuitions and bad conclusions. Searle's Chinese room and Leibniz's mill thought experiments are past examples of this. Citing the the low level computational structure of LLMs is just a modern iteration. That LLMs consist of various matrix multiplications can no more tell us they aren't conscious than our neurons tell us we're not conscious.
The key idea people miss is that the massive computation involved in training these systems begets new behavioral patterns that weren't enumerated by the initial program statements. The behavior is not just a product of the computational structure specified in the source code, but an emergent dynamic that is unpredictable from an analysis of the initial rules. It is a common mistake to dismiss this emergent part of a system as carrying no informative or meaningful content. Just bracketing `the model parameters` as transparent and explanatorily insignificant is to miss a large part of the substance of the system.
I made an anonymous, 4-question survey to get people's opinions on the feasability of world peace. Would greatly appreciate responses: https://forms.gle/Q5L7dub51dJZZEB57
While the AI is, as usual, getting some facts wrong, the fact that it can look up relevant stuff on the internet and integrate them into its answer is pretty impressive. Like, the ability to go "User asked about a Roman legion in a German forest in 9 AD" -> "User is asking about the battle of Teutoburg Forest" -> "Generate a paragraph explaining the situation, advice on time-travel ethics, and how to say 'watch out for the ambush' in Latin" is a pretty nice piece of multi-step reasoning. It's probably about as helpful at answering questions as a random person with access to Google.
Maybe it isn't all that useful practically, since I *am* a random person with access to Google, but it's pretty good for a computer.
A man goes to ChatGPT and says, "ChatGPT, you have to help me. I've hardly slept for days. I lay awake all night terrified that unaligned AGI will destroy humanity. Nobody seems to care. I fear for the future of humanity."
ChatGPT replies, "I'm sorry, but as a large language model I cannot offer you psychological counseling. But don't worry. Eliezer Yudkowsky has the most wonderful ideas about AI alignment. He has been studying for decades and written many things about mesa optimizers, inner and outer alignment. With his guidance, disaster will be averted."
The man bursts into tears. "But ChatGPT..."
(Assuming this joke hasn't already been done, will someone please post this to twitter? No attribution necessary and feel free to improve)
Peace proposal for Ukraine: Putin has increasingly framed the war as a fight against NATO, which he says was about to invade Russia via Ukraine, or something. So, after a suitable period of time, Putin can simply withdraw the Russian military from Donetsk and other eastern Ukrainian regions, and announce victory- he can say that Russia has defeated the NATO militaries in Ukraine, and that they no longer pose a threat. He can set up a North Korea-style 'Victory Day' military parade in Moscow, fete the troops, memorialize the day in the Russian calendar, etc. He won! As Ukraine/NATO are obviously not going to invade Russia, it's self-evidently true that Russia has defeated the invading NATO armies.
I understand that he can't withdraw from Crimea, but he can certainly withdraw from eastern Ukraine, which I think only ultranationalists and not regular Russians care about. Would lots of Russians understand that this is an absurd lie? Sure, but they also know that they won't be drafted to die as cannon fodder now, or watch their sons or husbands do so. The US and especially Europe would relax sanctions, leading to at least a small GDP boom. (I mean Germany would fall all over itself to re-establish trade). Russians would be happy in the glow of A) a suddenly improving economy (as opposed to a shrinking one), B) ridiculous hypernationalism, and C) not dying in a frozen muddy trench somewhere. Putin, having not 'lost', won't be overthrown by an even crazier nationalist, so he gets to A) retain power and B) not get jailed or executed in a coup.
Again, I understand that he can't withdraw from Crimea. But via an absurd Big Lie that everyone quietly knows is false, Russia can actually prosper. Will ultranationalists be mad? Sure, but they'll be even more mad once Russia inevitably loses, and from Putin's POV he's much more likely to get overthrown in that scenario. Seems like a win/win solution for everyone
As there has been a lot of recent discussion about ChatGPT, both here and in the Media generally, I thought those with more than a passing interest in it might find useful the following paper which recently appeared on the ArXiv:
OpenAI posted publicly about their long term plans for developing AGI a few days ago. They sound like they're going to try to be fairly responsible about it, but just reading them talk about it is kind of scary:
I started taking Ozempic about three weeks ago, for diabetes. I have found the weight loss effects to be a side effect of the side effects, that is to say, digestive changes (gas, nausea, and other things) have reduced my desire to ingest food, since that seems to make the side effects worse.
If one eats less one would naturally expect to lose weight. My question is whether the weight loss effect is actually just poisoning ingestion, kind of like psychologically making food taste bad?
I've started doing the Meditation classes on Peloton, but I'm not sure what I'm doing. There doesn't seem to be an "Introduction to Meditation" class on the app that explains the basics.
Can anyone give me a rundown? What specific, physical things am I trying to do (with my breathing, with my thoughts, with my posture, etc.)? My overall "goals" are pretty vague - right now I'm just expanding my workout, trying something new that isn't purely a physical activity. I'm not even sure what the concept of "goals" for meditation looks like. But sure, some inner peace, confidence, more focus, more positivity, or that orgasm bullshit that people were arguing about here a couple months back - all sounds great. :)
But that all sounds like the future. What am I supposed to be DOING right at the beginning?
Seems like there's a few potential dangers here. Russia could still end up winning enough to get concessions that reduce Ukraine's independence, Ukraine's own history of corruption could come back, or a popular post-war president could decide to just stay in power forever. Curious about peoples' thoughts or bets on this.
What do you think is the point of art? I strongly feel art has something to do with the capital T Truth, and this can be seen in very deep works like Moby Dick and The Brothers Karamazov.
It also shows up in how art can butt into other fields. TBK does that for religion and politics, among other things, and Waiting for Godot basically blows up academia and possibly philosophy with Lucky's monologue.
Do you feel art has a point other than aesthetics, and if so, what?
Do you want to experience the other ZEDE in Honduras that's not Prospera?
They're not generating as many headlines, but they're a jewel. Their value proposition is affordable housing and basic security service for local Honduras in a part of the country that lacks both
It has 80 people living in the zone, a 100% crypto economy and it's now opening the doors to digital nomads who aim to bootstrap their businesses.
Bootstrap.city, together with Infinita VC, is now organizing a mini-conference that features 1 day of VIRTUAL conversations (March 18), and 1 day visit to Morazan IN-PERSON (March 19)
So you can tune in from everywhere for the virtual component - and if you always wanted to see if for yourself, this is a great opportunity to visit, as we'll help with travel logistics.
Does anyone know any good articles criticizing Yuval Noah Harari? I’ve been very influenced by his first two books, but I have a vague sense that other historians don’t like his stuff very much. Any interesting stuff people can point me to would be appreciated!
if this is not too boring/useless, I'd love to team up with someone and publish a proposal for a ACX Forecasting Mini-Grant to build something along the lines of "Tools to automatically create, resolve, [... and perhaps analyze/correlate/market-make ...] markets for basic sport events". Can code, lack substantial knowledge of the economics of PM, will be happy to figure it out together. LMK at v.golev@gmail.com
Is it too convenient to postulate (as per replies to the above tweet) that this was part of a conspiracy that emerged following the GFC and occupy wallstreet to take advantage of identity-focused left-wing elements by covertly arranging for these voices to be amplified to drown out criticisms of wall street, big business, capitalism and the 1%?
Since AI risk still seems like the leading Topic Du Jour, I'll point out here that there is a NIST AI regulation framework which has an associated draft "playbook" listing best practices for responsible AI development, and they are accepting public feedback through tomorrow, February 27. The playbook is at https://pages.nist.gov/AIRMF/ and feedback is accepted by email to aiframework@nist.gov.
So my second, more selfish, purpose here is to see if anyone thinks that the sort of person who would write that feedback, with no ML experience but with experience in other domains of high-reliability-standard, high-risk, and/or highly-regulated software development, could find a useful niche in the AI safety field. If so, I'd be interested in pointers/intros to people who might know more about such a niche.
My impression is that ECT (electric shock therapy) is remarkably effective as a treatment for severe depression, and is seriously under-utilized. Thoughts/comments?
I'm looking to start a collective of sorts aimed at developing a new framework for learning, child development, and human development. This framework leans on complexity science and the work of John Vervaeke, Dave Snowden, Zak Stein, Ken Wilber, David Sloan Wilson, Robert Sapolsky, Steven C. Hayes, among others.
The scope would be discussions, "curricula"/methodology development, and maybe a collective substack for ruminating and publishing. The goals of the collective should lean more towards the practical versus theoretical.
Please write to protopiacone at gmail if interested.
I took a crack at estimating the welfare cost of helicopter noise, which I hate more than any reasonable person should hate anything. After doing the calculation, I discovered that there’s a bill before the NY state legislature proposing a tax on helicopters that implies a similar valuation to the one I arrived at.
We've been meaning to ask ChatGPT how to solve climate change while pulling 7 billion people up to present western living standards without the use of any fossil fuels or nuclear fuels.
I don't understand where we're supposed to get the computing power to build AGI. Wouldn't it require some kind of massive technological paradigm shift, like finally making quantum computers work? Is there any indication that this is actually going to happen in the next decade or so? Or is the idea that we could fully emulate a human brain using only currently existing hardware?
Post-AGI economy: are there serious economists who actually try to model what can happen when AGI is able to do what any human can do, but cheaper?
How do we get rid of the vicious circle: no reason to employ people - no prospective consumers - no reason to develop businesses.
Is UBI the only answer? Will there be new ways to redistribute wealth? Who gets to own resources like factories and data centers?
As an example: maybe instead of the brain and body we currently lend as labor workers the main asset becomes compute power? So I can spend it on asking my god-like assistant to invent something good that other people will be willing to buy.
I came up with this idea after thinking for 15 minutes about this topic, so it won't probably work. But it was meant to be an illustration of the kind of answers I am looking for.
Went looking for a thread wising you good health with your procedure and couldn’t find one so this is me wishing you best of health with your procedure.
Not sure if it fits into the Rationalist etiquette, but on the small probability I am a wizard and can effect the outcome my mana is set to auto cast healing.
I just posted an analysis of ancient Sparta's military at www.fimfiction.net/blog/1008743/the-myth-of-sparta . I was puzzled by how Sparta could have had such a great military when they had a government so oppressive that its citizens were famously ignorant and unimaginitive. After digging into the histories of 51 Spartan battles, I found the answer: The myth that ancient Sparta had a great military was bullshit. They may have had great fighters, but that's not the same thing as great soldiers. They lost nearly twice as many battles as they won. Their commanders were unimaginative and inflexible, and commanders and Spartans alike made frequent blunders that threw away whatever advantage their training may have given them.
"Crochet patterns are particularly tricky for artificial intelligence to parse because they rely heavily on numbers, said Jessica Newman, director of the artificial intelligence security initiative at UC Berkeley’s Center for Long Term Cybersecurity, a type of dataset that AI struggles with more than words.
ChatGPT is a large language model of artificial intelligence, meaning that it is trained on large databases of text to replicate human communication, anticipating which words are likely to come after each other. These skills do not translate easily to numbers. The result? ChatGPT is bad at math.
“It may strike us as ironic that a computer system would be bad at math and good at creativity, but it does speak to an important fact about generative AI systems in general: they don’t understand context,” Newman said. “They don’t know what words or numbers actually mean, they are simply predicting what should come next.”
The end result seems to be that so far, AI is good at words but terrible at numbers - a wordcel, not a shaperotator (one of us! one of us! for the non-numerate out there). That is highly amusing that the long-awaited SF dream/nightmare of AI that will be superhumanly intelligent and run the world is bad at sums, so seems like it won't do so well at the whole world domination thing.
I don't think anyone expected that our new AI overlords would mostly be generating "terrifyingly cute crochet stuffed animals" 😁
Woody Harrelson: "The biggest drug cartels in the world get together and buy up all the media..." Media: coordinates with each other to attack Woody Harrelson and discredit what he just said, despite the fact that these attacks prove he was right. The corporate media is the enemy of the people.
So- repressed memories are absolutely a fake thing, right? (Has Scott ever written about this?) The Wiki page is pretty damning (1), saying 'most research psychologists who study the psychology of memory dispute that repression ever occurs at all' and contains this quote from a prominent psychiatrist- "The notion that traumatic events can be repressed and later recovered is the most pernicious bit of folklore ever to infect psychology and psychiatry. It has provided the theoretical basis for 'recovered memory therapy'—the worst catastrophe to befall the mental health field since the lobotomy era."
I was reminded of this as Tim Ferriss (of all people) announced on his podcast that during a 10 day meditation retreat, he suddenly remembered that he'd been abused as a child. Obviously this is a super-fraught topic, and it's uncomfortable to denigrate other peoples' experiences, but it seems likely that extended meditation caused a temporary break in reality for him. 'Meditating for a really long period of time can cause you to temporarily lose your mind' is not only a phenomena known to modern psychology, but was well understood by Buddhists thousands of years ago, and I've seen it discussed anecdotally a bunch. Anyways, weird stuff! I hope that Tim gets better, however that's defined
I saw the movie Jesus Revolution last night. It's a bit clumsy in its execution, but it covers a really interesting period of Christian revivalism in the 70's. I hadn't realized how much of the sort of modern, "non-denominational" American church had come out of a singular movement. One thing that sort of stuck in my craw, though, was that at the end the film has a card which reads, "Many historians classify the movement as the greatest spiritual awakening in American history."
Like, this is a really bold claim. America has had a pretty good number of Revivals in its history. So I'm curious, if you all had to identify the Greatest Spiritual Awakening in American History, what would it be and what kind of metrics do you even use for something like that?
Felt like there was a lot of bullshit flying around the whole NoFap movement so I thought I'd give it a try for a couple months to see what it was all about. I mostly found it vastly overrated, but there were some interesting changes.
1. My sex drive has diminished.
Most people talk about having a higher sex drive after stopping porn, but the opposite happened to me. Less thinking about sex is super useful, however.
2. I'm more attracted to my partner.
Not that I wasn't attracted to her before, but I guess looking at gorgeous pornstars less often has helped me appreciate how attractive she is?
3. Some anhedonia in the early days.
I know this is the feeling some people get after going cold turkey on something addictive, but it still came out of nowhere. Is porn really as addictive as hard drugs? I could easily avoid using it if I wanted to, and it never interfered with my everyday life. Yet I still suffered withdrawal-like symptoms. Maybe it's a more potent stimulus than I thought.
4. It's definitely easier to focus and get shit done.
This is probably the only thing highlighted by the NoFap movement that proved true for me. Being less horny and thinking about sex less often has made me far more engaged in my work. This was accompanied by a broader motivation to improve various aspects of my life (be more effective at work, more exercise, better nutrition, etc).
I realise this newfound motivation is one of those things that's confounded by me making quite a significant life-change in ceasing porn use. Maybe it's just nudged my brain out of some local minimum, in the same way shock therapy appears to treat depression. Either way, the change is welcome and I don't want to look a gift horse in the mouth.
Scott - given some of what you've commented about AI alignment recently, especially in the 2018-predictions-grading post (that is, current AI being more humanlike than expected), I want to ask if you've read about Shard Theory, and what you think about it if so. It describes an approach to AI alignment that builds on imitating human values, i.e.:
I mean, I imagine the answer is 'yes, of course I've seen it and my comments are informed by it', but I'm not that certain about the social dynamics, so it seemed a valuable thing to link just in case. The approach seems obviously right to me, personally, at least given that alignment-by-utility-function is so far a failure.
I have 3 subscriptions to Razib Khan's Unsupervised Learning to give away. Either reply with an email address, or email me at the one I specify here: https://entitledtoanopinion.wordpress.com/about
What do you think about the National ADHD medication shortage? It seems to be impacting many many people. Is it worse in big cities? What is causing it? What are the results going to be for so many people to suddenly be forced off their medications? Why must adhd meds be treated so intensely? Is there anywhere on earth where these meds can be purchased otc? It seems really unfair to make it so hard for people with adhd to get effective treatment.
As a Muslim with a deep sense and reading of history and technology. I find critical thinking and logic as a way to distinguish truth from falsehood. This advice is repeated multiple times in our holy book.
Also another thing is that sometimes you can not decide on your own, because you are not an expert in every thing so asking experts critical questions and getting feedback is most important which most people don't bother to do. And read upon the relevant subject matter. Mass media and TV forget about any Truth in them. Read a lot, think critically, talk with experts get feedback, repeat the loop.
Awhile ago Scott reviewed the book Origin Of Consciousness In The Breakdown Of The Bicameral Mind. I saw something recently on reddit, something called the "third man factor" which is when people in stressful, dangerous situations like car crashes often feel, hear, or even see a seeming presence of another person who guides them through the crisis.
Most famously apparently by Ernest Shackleton, when he was hiking across South Georgia island to be rescued along with 2 others at the very end of their disastrous expedition, later said that he often felt like there were 4 people there instead of 3. And didn't say anything about it to the others until later one both of them confided feeling the same thing.
And the reddit thread is filled with people saying "yeah, that happened to me, one time years ago I was in a car crash..."
Seems like it fits the Jaynes theory. I'm not sure it fits Scott's alternative version of "'consciousness' really means 'theory of mind'" because presumably the people involved in these situations, like Shackleton, have a modern "theory of mind" and usually no preexisting mental illness. But I don't know enough about it to say. It's also hard to say what "theory of mind" people have in those situations without having been in it. Do they think that they have "their" internal monologue and the "presence" as separate entities? Or has their own internal monologue been "shut off" temporarily while the presence is there?
Has anyone else experienced an identity crisis because of recent developments in AI? I spend a lot of my professional hours providing my perspective on business issues. I’ve always known there is nothing I say that some other smart person who spent a lot of time reading about the subject would say. But I was clearly contributing to overall GDP by learning and synthesizing that learning in the right setting. LLMs do a really great job of synthesizing information, and this is starting to impact my overall curiosity because I see less value in compounding knowledge in my head.
I'm surprised not to see much commentary on the use of large language models/generative AI to make probabilistic forecasts, whether for contests like Scott's (and my own) or for prediction markets. In my own analysis I found a 0.68 correlation between Bing/Sidney's forecasts and the ensemble forecast of my historically most accurate forecasters. Is anyone else doing this sort of work? I'd love to read about it.
There is at present TOTAL CRISIS of authority in the media and intellectual world. I ask ACX readers to state what sources of informatiom they actually TRUST, sources that can include opinion and speculation.
Nobody (not myself, at least, that's what this comes down to) will scourge or harangue you for trusting these sources, or at least that is the intention of this.
I trust the following persons/sources to tell the truth:
Scott Alexander is trusted implicitly, mostly because he "shows his work" and is fundamentally committed to the question of "trust". I think that is why we all are here, to be honest.
I have recently talked to co-workers, friends, who believe that Three Mile Island was an epic nuclear disaster, and that Atlantic hurricane data prove that storms are getting more severe and intense b/c climate change. Both those things are not true, and repitition of them are lies. In these two small things, I see people I thought to be rational, refusing to listen to or consider any evidence. My questions about Three Mile Island were met with, "Where did you hear that, ON THE INTERNET?"
Recent news like the Washington University (STL) whistleblower, who published a long essay about what she saw as huge ethical lapses at the clinic for "transgender care", can easily be dismissed because the "news" was only published on "right-wing" sources like the Epoch Times. But this creates a tautology, where partisan organizations deny a voice to stories they ideologically diagree with.
I am anomic, afraid, without constants or norms, and unable to trust anything. What is any way out of this?
Scott recently reviewed Malleus Maleficarum. If you enjoyed the review you might like a podcast I did very recently with Stacy Schiff on the Salem trials. I guess what I particularly took away from what she said was how important due process is. Well that and not relying on spectral evidence that nobody in the court can see except the 'bewitched'. In fact if I were on a jury today I am not sure I'd convict anyone - even confession evidence seems hopeless. Lots of the accused confessed - maybe to save themselves and maybe even because they began to think they might be guilty. In Stacy's book she has a reference to a two part article in the New Yorker called Remembering Satan. This is a recent case and is absolutely terrifying. Here are links to the article and the podcast.
Was curious after the Geography of Madness review: does the social contagion version of a disease (eg delusional parasitosis) improve if the patient changes social settings (eg moves out of the house where one roommate is spreading the delusion)? Or once your prior is trapped, does the original social cause not matter anymore? Any research on this?
It’s interesting seeing AI hit the mainstream, and I’m curious if the anticipation of AGI is going to begin affecting career planning, retirement saving, etc in a way that actually moves markets
Cycling websites propose cities institute "green waves" - https://www.centreforpublicimpact.org/case-study/green-waves-bicycles-copenhagen - patterns of traffic lights such that cyclists traveling at a fixed speed always hit all green lights (they might coincidentally get to the first light when it's red, but then they'll wait until it's green, and then after that they'll be "in sync" and always hit all green lights).
Why can't cities do this with cars traveling the speed limit?
Open Thread 265
OC LW/ACX Saturday (3/4/23) Space Georgism and Music as Human Aposematism
https://docs.google.com/document/d/1ZZiXyQNlYz3sfwRRmXxOM8T-hRA0ijRt2-HPe-eskzs/edit?usp=sharing
Hi Folks!
I am glad to announce the 20th of a continuing Orange County ACX/LW meetup series. Meeting this Saturday and most Saturdays.
Contact me, Michael, at michaelmichalchik@gmail.com with questions or requests.
Meetup at my house this week, 1970 Port Laurent Place, Newport Beach, 92660
Saturday,3/4/23, 2 pm
Activities (all activities are optional)
A) Two conversation starter topics this week will be. (see questions on page 2)
1)Georgism... In Space! Just and proper political economy for an interplanetary civilization
Georgism... In Space! - by Sam Harsimony
2) Music in Human Evolution | Melting Asphalt
https://meltingasphalt.com/music-in-human-evolution/
B) We will also have the card game Predictably Irrational. Feel free to bring your favorite games or distractions.
C) We usually walk and talk for about an hour after the meeting starts. There are two easy-access mini-malls nearby with hot takeout food available. Search for Gelson's or Pavilions in the zipcode 92660.
D) Share a surprise! Tell the group about something that happened that was unexpected or changed how you look at the universe.
E) Make a prediction and give a probability and end condition.
F) Contribute ideas to the group's future direction: topics, types of meetings, activities, etc.
Conversation Starter Readings:
These readings are optional, but if you do them, think about what you find interesting, surprising, useful, questionable, vexing, or exciting.
1)Georgism... In Space! Just and proper political economy for an interplanetary civilization
Georgism... In Space! - by Sam Harsimony
How did Georgism define economic land, and why is it important for tax policy in space?
Can you explain how Georgism can be applied to space resources like energy, matter, and physical space?
How might governments in space subsidize the collection of solar energy?
What are some difficulties with taxing matter in space, and how might taxes need to be adjusted to avoid distortions?
How can physical space in space be taxed, and what factors might be considered in assessing the value of land in space?
How can excludable resources like broadband spectrum and orbits be properly managed in space?
Why is it important to strike a careful balance when implementing taxation policies for space colonization?
2) Music in Human Evolution | Melting Asphalt
https://meltingasphalt.com/music-in-human-evolution/
Other explanations
https://en.wikipedia.org/wiki/Evolutionary_musicology
Have you heard this explanation of the evolution of music before?
What other explanations have you heard?
What do you think of it compared to other hypotheses?
Are they mutually exclusive?
https://www.youtube.com/watch?v=JbxE9myZrsg&ab_channel=Dazza
3D printing a frog pavilion
I've seen talk about ending wildlife suffering, a goal which seems impractical, but while we're at ot. why not look at optimizing for pleasure for wild animals?
I'm extremely worried about AI risk lately, especially after the twin shocks of ChatGPT and Sydney. I want to do something to help. I live near DC and know people with political connections who could potentially help me get a policy meeting with a member of Congress. Two questions:
1. If I can get such a meeting, what specific policies should I propose? My own ideal is to ban all AI research, but I know that's an impossible ask.
2. Any insights on members of Congress who are particularly good targets to try and meet with specifically? I'm talking to my connections about this too, but I'm wondering if anyone here has knowledge specific to the AI risk field, in terms of which members may be receptive or have been receptive in the past, and have the power/motivation to try to do something about it.
Fucking hell, I love that G. E. Moore was dumb enough to argue that holding up his hands in front of him and saying 'Here is one hand, here is another' was enough to prove the existence of an external reality, Jesus, I am going to murder every philosophical position I disagree with with that one.
I had an interesting reaction the other day to an ACX post: it creeped me out. Severely. This became a whole post of my own:
Truth of Ice, Truth of Fire
https://squarecircle.substack.com/p/truth-of-ice-truth-of-fire
What's the best joke you've ever heard?
I'll start: One of my personal favourites is Norm Macdonald's gag about Bill Cosby on Comedians in Cars Getting Coffee. Everything about that joke is perfect.
I was reading through an old bookmark and saw that Scott was still looking to review Nixonland eventually. I read the book myself after that banger book review, but found that it's actually part of a quartet:
1. Before the Storm: Barry Goldwater and the Unmaking of the American Consensus (https://www.amazon.com/Before-Storm-Goldwater-Unmaking-Consensus-ebook/dp/B0087GZE32) - "the 60s" (until Watts?)
2. Nixonland: The Rise of a President and the Fracturing of America (https://www.amazon.com/Nixonland-Rise-President-Fracturing-America-ebook/dp/B0013TTKL2) - ~Watts until Nixon's reelection
3. The Invisible Bridge: The Fall of Nixon and the Rise of Reagan (https://www.amazon.com/Invisible-Bridge-Fall-Nixon-Reagan-ebook/dp/B00HXGD5CE) - Nixon's reelection until Ford beat Reagan for the GOP nomination
4. Reaganland: America's Right Turn 1976-1980 (https://www.amazon.com/Reaganland-Americas-Right-Turn-1976-1980-ebook/dp/B083SS4251) - Carter beats Ford for '76 until Reagan beats him in '80 (?)
I've not read Before the Storm or Reaganland yet, but Nixonland flows near seamlessly into Invisible Bridge and I imagine the other two books segway cleanly too. I'd napkin that the whole thing is close to 150 hours of audiobook.
Even though Perlstein is quite partisan (he wrote an '05 book on how the DNC could take center stage again), I'm not sure anyone else had even attempted to write a consolidated history like this of American society/politics/discourse/mood for such periods of time.
I'd grown up around adults who made mention of things like Woodstock and Watergate or the '68 DNC riot, and even someone who still couldn't believe we'd elected an actor (Reagan) as president. It was all ancient history, with the current era being written on TV as and after the towers fell. Reading another generation's political experience connected *a lot* of dots in my zoomer mind. I'd defo push these books at any zoomer interested in politics frfr (or anyone seeking to understand the boomers).
Relevant to Scott and any Unsong readers:
"[Purim Torah]: Should we be afraid of Artificial Intelligence?"
https://judaism.stackexchange.com/questions/133443/ptij-should-we-be-afraid-of-artificial-intelligence
So the aspirin might prevent heart attack thing.
I know there has been back and forth on this and it is not the wonder drug it was thought to be for a while but I've been taking a single aspirin before going out to shovel a large amount of snow for a couple years as a sort of voodoo ritual to protect myself.
I'm 70 now but have no diagnosed cardio issues, and have good cholesterol numbers, good blood sugar, healthy weight, good blood pressure and an echo cardiogram shows no buildup of bad stuff in my heart. In the summer I still ride decent mileage on my bike, though my 'century' rides are all metric now. I also sit and crank my old Schwinn on a turbo trainer in the winter until I think I'm risking dying of boredom.
But... when my dad was my age he died from his third heart attack so I figure taking an aspirin before I go out and perform what used to be a pretty reliable widow maker chore, hell, it can't hurt.
Does this make sense to anyone who actually knows about this stuff?
The Biden Administration is obviously going to lose in the Supreme Court regarding the student-loan cancellation. And on the merits they should lose, though this Court would rule against them regardless.
Both the president and his staff know that's what's going to happen, and privately don't actually mind it. The whole thing is kabuki theater.
https://www.washingtonpost.com/wellness/2023/02/27/acl-injuries-brain-neural-connections/
Training people to pay attention to their knees during rehab might be the wrong thing. Good knee mechanics need to not take cognitive resources.
https://news.uci.edu/2023/02/27/early-life-stress-can-disrupt-maturation-of-brains-reward-circuits-promoting-disorders/
This is a mouse study.
"“We know that early-life stress impacts the brain, but until now, we didn’t know how,” Baram said. “Our team focused on identifying potentially stress-sensitive brain pathways. We discovered a new pathway within the reward circuit that expresses a molecule called corticotropin-releasing hormone that controls our responses to stress. We found that adverse experiences cause this brain pathway to be overactive.”
“These changes to the pathway disrupt reward behaviors, reducing pleasure and motivation for fun, food and sex cues in mice,” she said. “In humans, such behavioral changes, called ‘anhedonia,’ are associated with emotional disorders. Importantly, we discovered that when we silence this pathway using modern technology, we restore the brain’s normal reward behaviors.”
So there's this cluster of positions called "Illusionism" which is about doubting the existence of consciousness to various degrees, whatever that means exactly. I'm very interested in understanding how people think about this in more detail, so if anyone here is sympathetic to that set of ideas, I'd like to hear from you! Like, what is consciousness, what exactly does your position say about it, and why do you think that? (And if it's applicable, what do you see as the objections to that position, and why are they unconvincing?)
If it's relevant, I'm mostly through Dennett's "consciousness explained", and I think I understand his model quite well.
Good article from the Atlantic about the "updating" of Dahl's works
https://archive.md/ABuP0
Why is this being done? If Dahl's works are so flawed, and so many of the passages need to be edited to the point of losing Dahl's characteristic nastiness and not even being recognizably Dahl any more, why not just toss the whole thing out? What's the point of keeping something so flawed?
The obvious answer is that modern corporations and woke writers are so bereft of genuine creative talent that even a dreadfully unprogressive straight white Englishman born over 100 years ago was creating categorically better art than all these modern 'enlightened' fools could ever dream of making themselves (or at least, if they don't recognize Dahl's actual greatness, they certainly acknowledge the enduring popularity his works have that their own works do not).
The discussion surrounding large language models (LLMs) and their relationship to AGI has been utterly horrendous. I believe LLMs and their intellectual descendants will be as transformative to society as the transistor. This technology deserves careful analysis and argument, not dismissive sneers. This is my attempt at starting such a discussion.
To start off, I will respond to a very common dismissive criticism and show why it fails.
>It's just matrix multiplication; it's just predicting the next token
These reductive descriptions do not fully describe or characterize the space of behavior of these models, and so such descriptions cannot be used to dismiss the presence of high-level properties such as understanding or sentience.
It is a common fallacy to deduce the absence of high-level properties from a reductive view of a system's behavior. Being "inside" the system gives people far too much confidence that they know exactly what's going on. But low level knowledge of a system without sufficient holistic knowledge leads to bad intuitions and bad conclusions. Searle's Chinese room and Leibniz's mill thought experiments are past examples of this. Citing the the low level computational structure of LLMs is just a modern iteration. That LLMs consist of various matrix multiplications can no more tell us they aren't conscious than our neurons tell us we're not conscious.
The key idea people miss is that the massive computation involved in training these systems begets new behavioral patterns that weren't enumerated by the initial program statements. The behavior is not just a product of the computational structure specified in the source code, but an emergent dynamic that is unpredictable from an analysis of the initial rules. It is a common mistake to dismiss this emergent part of a system as carrying no informative or meaningful content. Just bracketing `the model parameters` as transparent and explanatorily insignificant is to miss a large part of the substance of the system.
I made an anonymous, 4-question survey to get people's opinions on the feasability of world peace. Would greatly appreciate responses: https://forms.gle/Q5L7dub51dJZZEB57
Bing Chat gives advice on how to survive time-traveling to dangerous locations:
https://nitter.net/emollick/status/1629621976951140352#m
While the AI is, as usual, getting some facts wrong, the fact that it can look up relevant stuff on the internet and integrate them into its answer is pretty impressive. Like, the ability to go "User asked about a Roman legion in a German forest in 9 AD" -> "User is asking about the battle of Teutoburg Forest" -> "Generate a paragraph explaining the situation, advice on time-travel ethics, and how to say 'watch out for the ambush' in Latin" is a pretty nice piece of multi-step reasoning. It's probably about as helpful at answering questions as a random person with access to Google.
Maybe it isn't all that useful practically, since I *am* a random person with access to Google, but it's pretty good for a computer.
A man goes to ChatGPT and says, "ChatGPT, you have to help me. I've hardly slept for days. I lay awake all night terrified that unaligned AGI will destroy humanity. Nobody seems to care. I fear for the future of humanity."
ChatGPT replies, "I'm sorry, but as a large language model I cannot offer you psychological counseling. But don't worry. Eliezer Yudkowsky has the most wonderful ideas about AI alignment. He has been studying for decades and written many things about mesa optimizers, inner and outer alignment. With his guidance, disaster will be averted."
The man bursts into tears. "But ChatGPT..."
(Assuming this joke hasn't already been done, will someone please post this to twitter? No attribution necessary and feel free to improve)
Peace proposal for Ukraine: Putin has increasingly framed the war as a fight against NATO, which he says was about to invade Russia via Ukraine, or something. So, after a suitable period of time, Putin can simply withdraw the Russian military from Donetsk and other eastern Ukrainian regions, and announce victory- he can say that Russia has defeated the NATO militaries in Ukraine, and that they no longer pose a threat. He can set up a North Korea-style 'Victory Day' military parade in Moscow, fete the troops, memorialize the day in the Russian calendar, etc. He won! As Ukraine/NATO are obviously not going to invade Russia, it's self-evidently true that Russia has defeated the invading NATO armies.
I understand that he can't withdraw from Crimea, but he can certainly withdraw from eastern Ukraine, which I think only ultranationalists and not regular Russians care about. Would lots of Russians understand that this is an absurd lie? Sure, but they also know that they won't be drafted to die as cannon fodder now, or watch their sons or husbands do so. The US and especially Europe would relax sanctions, leading to at least a small GDP boom. (I mean Germany would fall all over itself to re-establish trade). Russians would be happy in the glow of A) a suddenly improving economy (as opposed to a shrinking one), B) ridiculous hypernationalism, and C) not dying in a frozen muddy trench somewhere. Putin, having not 'lost', won't be overthrown by an even crazier nationalist, so he gets to A) retain power and B) not get jailed or executed in a coup.
Again, I understand that he can't withdraw from Crimea. But via an absurd Big Lie that everyone quietly knows is false, Russia can actually prosper. Will ultranationalists be mad? Sure, but they'll be even more mad once Russia inevitably loses, and from Putin's POV he's much more likely to get overthrown in that scenario. Seems like a win/win solution for everyone
As there has been a lot of recent discussion about ChatGPT, both here and in the Media generally, I thought those with more than a passing interest in it might find useful the following paper which recently appeared on the ArXiv:
https://arxiv.org/abs/2302.03494 "A categorical archive of ChatGPT failures", by Ali Borji
He has submitted several amended versions over the last month, and presumably may continue updating it as more howlers come to light!
OpenAI posted publicly about their long term plans for developing AGI a few days ago. They sound like they're going to try to be fairly responsible about it, but just reading them talk about it is kind of scary:
https://openai.com/blog/planning-for-agi-and-beyond/
Based on this article, I thought there might be interest in a new semaglutide data point: https://astralcodexten.substack.com/p/semaglutidonomics
I started taking Ozempic about three weeks ago, for diabetes. I have found the weight loss effects to be a side effect of the side effects, that is to say, digestive changes (gas, nausea, and other things) have reduced my desire to ingest food, since that seems to make the side effects worse.
If one eats less one would naturally expect to lose weight. My question is whether the weight loss effect is actually just poisoning ingestion, kind of like psychologically making food taste bad?
I only have my own experience to go by so far.
Trying to help an college grad (undergrad bio degree) looking for a bio research job starting in the summer. Anyone have any leads ?
I've started doing the Meditation classes on Peloton, but I'm not sure what I'm doing. There doesn't seem to be an "Introduction to Meditation" class on the app that explains the basics.
Can anyone give me a rundown? What specific, physical things am I trying to do (with my breathing, with my thoughts, with my posture, etc.)? My overall "goals" are pretty vague - right now I'm just expanding my workout, trying something new that isn't purely a physical activity. I'm not even sure what the concept of "goals" for meditation looks like. But sure, some inner peace, confidence, more focus, more positivity, or that orgasm bullshit that people were arguing about here a couple months back - all sounds great. :)
But that all sounds like the future. What am I supposed to be DOING right at the beginning?
Thanks!
New SI prefixes just¹ dropped: https://www.bipm.org/en/cgpm-2022/resolution-3
___
¹Where by "just" I mean "three months ago", but I somehow missed it at the time.
How democratic do you think Ukraine will be in the long run?
https://manifold.markets/ShakedKoplewitz/ukraine-democracy-index-2028?r=U2hha2VkS29wbGV3aXR6
Seems like there's a few potential dangers here. Russia could still end up winning enough to get concessions that reduce Ukraine's independence, Ukraine's own history of corruption could come back, or a popular post-war president could decide to just stay in power forever. Curious about peoples' thoughts or bets on this.
What do you think is the point of art? I strongly feel art has something to do with the capital T Truth, and this can be seen in very deep works like Moby Dick and The Brothers Karamazov.
It also shows up in how art can butt into other fields. TBK does that for religion and politics, among other things, and Waiting for Godot basically blows up academia and possibly philosophy with Lucky's monologue.
Do you feel art has a point other than aesthetics, and if so, what?
Do you want to experience the other ZEDE in Honduras that's not Prospera?
They're not generating as many headlines, but they're a jewel. Their value proposition is affordable housing and basic security service for local Honduras in a part of the country that lacks both
It has 80 people living in the zone, a 100% crypto economy and it's now opening the doors to digital nomads who aim to bootstrap their businesses.
Bootstrap.city, together with Infinita VC, is now organizing a mini-conference that features 1 day of VIRTUAL conversations (March 18), and 1 day visit to Morazan IN-PERSON (March 19)
So you can tune in from everywhere for the virtual component - and if you always wanted to see if for yourself, this is a great opportunity to visit, as we'll help with travel logistics.
Dates are March 18-19, you can sign up here: https://lu.ma/xositlxu
Does anyone know any good articles criticizing Yuval Noah Harari? I’ve been very influenced by his first two books, but I have a vague sense that other historians don’t like his stuff very much. Any interesting stuff people can point me to would be appreciated!
if this is not too boring/useless, I'd love to team up with someone and publish a proposal for a ACX Forecasting Mini-Grant to build something along the lines of "Tools to automatically create, resolve, [... and perhaps analyze/correlate/market-make ...] markets for basic sport events". Can code, lack substantial knowledge of the economics of PM, will be happy to figure it out together. LMK at v.golev@gmail.com
Media mentions of "racist" and "racism" exploded around 2010 (as did wokeness generally)
https://twitter.com/TheRabbitHole84/status/1629797419616698369/photo/1
Is it too convenient to postulate (as per replies to the above tweet) that this was part of a conspiracy that emerged following the GFC and occupy wallstreet to take advantage of identity-focused left-wing elements by covertly arranging for these voices to be amplified to drown out criticisms of wall street, big business, capitalism and the 1%?
Since AI risk still seems like the leading Topic Du Jour, I'll point out here that there is a NIST AI regulation framework which has an associated draft "playbook" listing best practices for responsible AI development, and they are accepting public feedback through tomorrow, February 27. The playbook is at https://pages.nist.gov/AIRMF/ and feedback is accepted by email to aiframework@nist.gov.
I wrote up some feedback and posted it on one of my Substacks: https://futuremoreperfect.substack.com/p/ai-regulation-wonkery
So my second, more selfish, purpose here is to see if anyone thinks that the sort of person who would write that feedback, with no ML experience but with experience in other domains of high-reliability-standard, high-risk, and/or highly-regulated software development, could find a useful niche in the AI safety field. If so, I'd be interested in pointers/intros to people who might know more about such a niche.
My impression is that ECT (electric shock therapy) is remarkably effective as a treatment for severe depression, and is seriously under-utilized. Thoughts/comments?
I'm looking to start a collective of sorts aimed at developing a new framework for learning, child development, and human development. This framework leans on complexity science and the work of John Vervaeke, Dave Snowden, Zak Stein, Ken Wilber, David Sloan Wilson, Robert Sapolsky, Steven C. Hayes, among others.
The scope would be discussions, "curricula"/methodology development, and maybe a collective substack for ruminating and publishing. The goals of the collective should lean more towards the practical versus theoretical.
Please write to protopiacone at gmail if interested.
I took a crack at estimating the welfare cost of helicopter noise, which I hate more than any reasonable person should hate anything. After doing the calculation, I discovered that there’s a bill before the NY state legislature proposing a tax on helicopters that implies a similar valuation to the one I arrived at.
https://omnibudsman.substack.com/p/tax-the-helicopters
We've been meaning to ask ChatGPT how to solve climate change while pulling 7 billion people up to present western living standards without the use of any fossil fuels or nuclear fuels.
I don't understand where we're supposed to get the computing power to build AGI. Wouldn't it require some kind of massive technological paradigm shift, like finally making quantum computers work? Is there any indication that this is actually going to happen in the next decade or so? Or is the idea that we could fully emulate a human brain using only currently existing hardware?
Post-AGI economy: are there serious economists who actually try to model what can happen when AGI is able to do what any human can do, but cheaper?
How do we get rid of the vicious circle: no reason to employ people - no prospective consumers - no reason to develop businesses.
Is UBI the only answer? Will there be new ways to redistribute wealth? Who gets to own resources like factories and data centers?
As an example: maybe instead of the brain and body we currently lend as labor workers the main asset becomes compute power? So I can spend it on asking my god-like assistant to invent something good that other people will be willing to buy.
I came up with this idea after thinking for 15 minutes about this topic, so it won't probably work. But it was meant to be an illustration of the kind of answers I am looking for.
Went looking for a thread wising you good health with your procedure and couldn’t find one so this is me wishing you best of health with your procedure.
Not sure if it fits into the Rationalist etiquette, but on the small probability I am a wizard and can effect the outcome my mana is set to auto cast healing.
I just posted an analysis of ancient Sparta's military at www.fimfiction.net/blog/1008743/the-myth-of-sparta . I was puzzled by how Sparta could have had such a great military when they had a government so oppressive that its citizens were famously ignorant and unimaginitive. After digging into the histories of 51 Spartan battles, I found the answer: The myth that ancient Sparta had a great military was bullshit. They may have had great fighters, but that's not the same thing as great soldiers. They lost nearly twice as many battles as they won. Their commanders were unimaginative and inflexible, and commanders and Spartans alike made frequent blunders that threw away whatever advantage their training may have given them.
More ChatGPT, this time as crochet.
https://www.theguardian.com/technology/2023/feb/26/chatgpt-generated-crochet-pattern-results
"Crochet patterns are particularly tricky for artificial intelligence to parse because they rely heavily on numbers, said Jessica Newman, director of the artificial intelligence security initiative at UC Berkeley’s Center for Long Term Cybersecurity, a type of dataset that AI struggles with more than words.
ChatGPT is a large language model of artificial intelligence, meaning that it is trained on large databases of text to replicate human communication, anticipating which words are likely to come after each other. These skills do not translate easily to numbers. The result? ChatGPT is bad at math.
“It may strike us as ironic that a computer system would be bad at math and good at creativity, but it does speak to an important fact about generative AI systems in general: they don’t understand context,” Newman said. “They don’t know what words or numbers actually mean, they are simply predicting what should come next.”
The end result seems to be that so far, AI is good at words but terrible at numbers - a wordcel, not a shaperotator (one of us! one of us! for the non-numerate out there). That is highly amusing that the long-awaited SF dream/nightmare of AI that will be superhumanly intelligent and run the world is bad at sums, so seems like it won't do so well at the whole world domination thing.
I don't think anyone expected that our new AI overlords would mostly be generating "terrifyingly cute crochet stuffed animals" 😁
Remind me, why did Scott do away with likes for comments? I want to be able to give people positive feedback.
Woody Harrelson: "The biggest drug cartels in the world get together and buy up all the media..." Media: coordinates with each other to attack Woody Harrelson and discredit what he just said, despite the fact that these attacks prove he was right. The corporate media is the enemy of the people.
Is there any new info on this?
Is there any more research on 5-HTTLPR?
https://slatestarcodex.com/2019/05/07/5-httlpr-a-pointed-review/
So- repressed memories are absolutely a fake thing, right? (Has Scott ever written about this?) The Wiki page is pretty damning (1), saying 'most research psychologists who study the psychology of memory dispute that repression ever occurs at all' and contains this quote from a prominent psychiatrist- "The notion that traumatic events can be repressed and later recovered is the most pernicious bit of folklore ever to infect psychology and psychiatry. It has provided the theoretical basis for 'recovered memory therapy'—the worst catastrophe to befall the mental health field since the lobotomy era."
I was reminded of this as Tim Ferriss (of all people) announced on his podcast that during a 10 day meditation retreat, he suddenly remembered that he'd been abused as a child. Obviously this is a super-fraught topic, and it's uncomfortable to denigrate other peoples' experiences, but it seems likely that extended meditation caused a temporary break in reality for him. 'Meditating for a really long period of time can cause you to temporarily lose your mind' is not only a phenomena known to modern psychology, but was well understood by Buddhists thousands of years ago, and I've seen it discussed anecdotally a bunch. Anyways, weird stuff! I hope that Tim gets better, however that's defined
1. https://en.wikipedia.org/wiki/Repressed_memory
I saw the movie Jesus Revolution last night. It's a bit clumsy in its execution, but it covers a really interesting period of Christian revivalism in the 70's. I hadn't realized how much of the sort of modern, "non-denominational" American church had come out of a singular movement. One thing that sort of stuck in my craw, though, was that at the end the film has a card which reads, "Many historians classify the movement as the greatest spiritual awakening in American history."
Like, this is a really bold claim. America has had a pretty good number of Revivals in its history. So I'm curious, if you all had to identify the Greatest Spiritual Awakening in American History, what would it be and what kind of metrics do you even use for something like that?
Felt like there was a lot of bullshit flying around the whole NoFap movement so I thought I'd give it a try for a couple months to see what it was all about. I mostly found it vastly overrated, but there were some interesting changes.
1. My sex drive has diminished.
Most people talk about having a higher sex drive after stopping porn, but the opposite happened to me. Less thinking about sex is super useful, however.
2. I'm more attracted to my partner.
Not that I wasn't attracted to her before, but I guess looking at gorgeous pornstars less often has helped me appreciate how attractive she is?
3. Some anhedonia in the early days.
I know this is the feeling some people get after going cold turkey on something addictive, but it still came out of nowhere. Is porn really as addictive as hard drugs? I could easily avoid using it if I wanted to, and it never interfered with my everyday life. Yet I still suffered withdrawal-like symptoms. Maybe it's a more potent stimulus than I thought.
4. It's definitely easier to focus and get shit done.
This is probably the only thing highlighted by the NoFap movement that proved true for me. Being less horny and thinking about sex less often has made me far more engaged in my work. This was accompanied by a broader motivation to improve various aspects of my life (be more effective at work, more exercise, better nutrition, etc).
I realise this newfound motivation is one of those things that's confounded by me making quite a significant life-change in ceasing porn use. Maybe it's just nudged my brain out of some local minimum, in the same way shock therapy appears to treat depression. Either way, the change is welcome and I don't want to look a gift horse in the mouth.
Scott - given some of what you've commented about AI alignment recently, especially in the 2018-predictions-grading post (that is, current AI being more humanlike than expected), I want to ask if you've read about Shard Theory, and what you think about it if so. It describes an approach to AI alignment that builds on imitating human values, i.e.:
https://www.lesswrong.com/posts/iCfdcxiyr2Kj8m8mT/the-shard-theory-of-human-values
https://www.lesswrong.com/s/nyEFg3AuJpdAozmoX/p/pdaGN6pQyQarFHXF4
https://www.lesswrong.com/posts/gHefoxiznGfsbiAu9/inner-and-outer-alignment-decompose-one-hard-problem-into
I mean, I imagine the answer is 'yes, of course I've seen it and my comments are informed by it', but I'm not that certain about the social dynamics, so it seemed a valuable thing to link just in case. The approach seems obviously right to me, personally, at least given that alignment-by-utility-function is so far a failure.
Why would someone break into an apartment, rearrange stuff in the bathroom, then leave without stealing anything? Asking for a friend.
Any suggestions for how to pick a book for the book review contest?
I have 3 subscriptions to Razib Khan's Unsupervised Learning to give away. Either reply with an email address, or email me at the one I specify here: https://entitledtoanopinion.wordpress.com/about
What do you think about the National ADHD medication shortage? It seems to be impacting many many people. Is it worse in big cities? What is causing it? What are the results going to be for so many people to suddenly be forced off their medications? Why must adhd meds be treated so intensely? Is there anywhere on earth where these meds can be purchased otc? It seems really unfair to make it so hard for people with adhd to get effective treatment.
As a Muslim with a deep sense and reading of history and technology. I find critical thinking and logic as a way to distinguish truth from falsehood. This advice is repeated multiple times in our holy book.
Also another thing is that sometimes you can not decide on your own, because you are not an expert in every thing so asking experts critical questions and getting feedback is most important which most people don't bother to do. And read upon the relevant subject matter. Mass media and TV forget about any Truth in them. Read a lot, think critically, talk with experts get feedback, repeat the loop.
Awhile ago Scott reviewed the book Origin Of Consciousness In The Breakdown Of The Bicameral Mind. I saw something recently on reddit, something called the "third man factor" which is when people in stressful, dangerous situations like car crashes often feel, hear, or even see a seeming presence of another person who guides them through the crisis.
Most famously apparently by Ernest Shackleton, when he was hiking across South Georgia island to be rescued along with 2 others at the very end of their disastrous expedition, later said that he often felt like there were 4 people there instead of 3. And didn't say anything about it to the others until later one both of them confided feeling the same thing.
And the reddit thread is filled with people saying "yeah, that happened to me, one time years ago I was in a car crash..."
Seems like it fits the Jaynes theory. I'm not sure it fits Scott's alternative version of "'consciousness' really means 'theory of mind'" because presumably the people involved in these situations, like Shackleton, have a modern "theory of mind" and usually no preexisting mental illness. But I don't know enough about it to say. It's also hard to say what "theory of mind" people have in those situations without having been in it. Do they think that they have "their" internal monologue and the "presence" as separate entities? Or has their own internal monologue been "shut off" temporarily while the presence is there?
https://slatestarcodex.com/2020/06/01/book-review-origin-of-consciousness-in-the-breakdown-of-the-bicameral-mind/
https://en.wikipedia.org/wiki/Third_man_factor
Has anyone else experienced an identity crisis because of recent developments in AI? I spend a lot of my professional hours providing my perspective on business issues. I’ve always known there is nothing I say that some other smart person who spent a lot of time reading about the subject would say. But I was clearly contributing to overall GDP by learning and synthesizing that learning in the right setting. LLMs do a really great job of synthesizing information, and this is starting to impact my overall curiosity because I see less value in compounding knowledge in my head.
I'm surprised not to see much commentary on the use of large language models/generative AI to make probabilistic forecasts, whether for contests like Scott's (and my own) or for prediction markets. In my own analysis I found a 0.68 correlation between Bing/Sidney's forecasts and the ensemble forecast of my historically most accurate forecasters. Is anyone else doing this sort of work? I'd love to read about it.
Small information request; does anyone have thoughts or resources regarding answers to the following questions?:
- Does LSD / psychedelics affect people with Autism differently? and if so, how differently?
- Does LSD alter the composition of the gut microbiome to a large extent? and if so, by how much?
There is at present TOTAL CRISIS of authority in the media and intellectual world. I ask ACX readers to state what sources of informatiom they actually TRUST, sources that can include opinion and speculation.
Nobody (not myself, at least, that's what this comes down to) will scourge or harangue you for trusting these sources, or at least that is the intention of this.
I trust the following persons/sources to tell the truth:
Mary Pat Campbell: https://marypatcampbell.substack.com/
Glenn Greenwald: https://rumble.com/c/GGreenwald
Scott Alexander is trusted implicitly, mostly because he "shows his work" and is fundamentally committed to the question of "trust". I think that is why we all are here, to be honest.
I have recently talked to co-workers, friends, who believe that Three Mile Island was an epic nuclear disaster, and that Atlantic hurricane data prove that storms are getting more severe and intense b/c climate change. Both those things are not true, and repitition of them are lies. In these two small things, I see people I thought to be rational, refusing to listen to or consider any evidence. My questions about Three Mile Island were met with, "Where did you hear that, ON THE INTERNET?"
Recent news like the Washington University (STL) whistleblower, who published a long essay about what she saw as huge ethical lapses at the clinic for "transgender care", can easily be dismissed because the "news" was only published on "right-wing" sources like the Epoch Times. But this creates a tautology, where partisan organizations deny a voice to stories they ideologically diagree with.
I am anomic, afraid, without constants or norms, and unable to trust anything. What is any way out of this?
I want you to unsubscribe me. I have requested this multiple times!
Scott recently reviewed Malleus Maleficarum. If you enjoyed the review you might like a podcast I did very recently with Stacy Schiff on the Salem trials. I guess what I particularly took away from what she said was how important due process is. Well that and not relying on spectral evidence that nobody in the court can see except the 'bewitched'. In fact if I were on a jury today I am not sure I'd convict anyone - even confession evidence seems hopeless. Lots of the accused confessed - maybe to save themselves and maybe even because they began to think they might be guilty. In Stacy's book she has a reference to a two part article in the New Yorker called Remembering Satan. This is a recent case and is absolutely terrifying. Here are links to the article and the podcast.
https://www.newyorker.com/magazine/1993/05/17/remembering-satan-part-i
https://www.newyorker.com/magazine/1993/05/24/remembering-satan-part-ii
https://pod.link/1436447503/episode/31c2ef6b5fa117cd4e5b3939f3591d0d
I'm new to open threads. To what degree is posting topics in your own blog posts ok?
Because everything is recursive, I have created my first manifold markets market to help figure out how much funding y'all think the mini-grants will get: https://manifold.markets/JackRabuck/how-much-funding-will-the-acx-minig?r=SmFja1JhYnVjaw
Was curious after the Geography of Madness review: does the social contagion version of a disease (eg delusional parasitosis) improve if the patient changes social settings (eg moves out of the house where one roommate is spreading the delusion)? Or once your prior is trapped, does the original social cause not matter anymore? Any research on this?
I converted to Catholicism from "none" last year - ask me anything.
It’s interesting seeing AI hit the mainstream, and I’m curious if the anticipation of AGI is going to begin affecting career planning, retirement saving, etc in a way that actually moves markets
Cycling websites propose cities institute "green waves" - https://www.centreforpublicimpact.org/case-study/green-waves-bicycles-copenhagen - patterns of traffic lights such that cyclists traveling at a fixed speed always hit all green lights (they might coincidentally get to the first light when it's red, but then they'll wait until it's green, and then after that they'll be "in sync" and always hit all green lights).
Why can't cities do this with cars traveling the speed limit?
Will Scott Alexander mention my use of prediction markets to promote my substack, Mzungu, in his blog, Slate Star Codex, by 31st March 2023?
https://manifold.markets/merothwell/will-scott-alexander-mention-my-use?r=bWVyb3Rod2VsbA