nim

Karma: 526

nim 22 Mar 2023 18:21 UTC
1 point
0
on: Deep Deceptiveness
How do we teach human children to not behave deceptively?

If we can do it with humans, why doesn’t that scale to AI? And if we can’t do it with humans, what factors keep it in check among humans, and how can those factors scale for AI?

nim 22 Mar 2023 16:58 UTC
1 point
0
on: Sun-following Garden Mirrors?
I am not aware of a heliostat designed for mirrors, but the problem sounds like it’s shaped similarly to the problems addressed by solar tracker systems for solar panels. I wonder if a system like that could be repurposed for your needs?

I’d be nervous about wind loads and breakage filling our own front garden with shards

I second the intuition against putting something up in the air that might shatter.

Would a driveway mirror work? I believe that they’re commonly made of shatter-resistant materials—shiny metal, or appropriate plastics.

Would a mylar-faced tarp work for prototyping? It might be ugly and it might be loud in the wind, but “space tarps” sold for survival situations are highly reflective on one side and relatively sturdy. A flexible reflector could be hoisted in good weather and furled in bad with the same basic technologies that have moved sailing ships around the world for centuries. Space blankets are too flimsy to hold up well on their own and would probably require additional support to not tear… although many types of space blanket can be ironed onto a sturdy fabric like a canvas drop cloth if you experiment a bit to find the right setting on your clothes iron.

Would a chromed metal panel, or even bright white metal roofing, work as a mirror without the shatter risk?

nim 18 Mar 2023 1:06 UTC
3 points
0
on: Downvote without reason?
Is it even a community goal for consensus to be “scientific”, whatever that means to you?

nim 17 Mar 2023 16:18 UTC
1 point
0
in reply to: jmh’s comment on: Are the LLM “intelligence” tests publicly available for humans to take?
Thank you! I didn’t see your first version of this, but your current version is helpful for the human-specific tests that they’re benchmarked on :)

[Question] Are the LLM “intelligence” tests publicly available for humans to take?

nim17 Mar 2023 0:09 UTC

6 points

8 comments1 min readLW link

nim 12 Mar 2023 18:51 UTC
1 point
0
in reply to: sudo -i’s comment on: Minor Life Optimization: Consider Ordering Your Food To-Go
Have you considered ordering catering for a “group” a couple times a week, and having your meals from the single catering order for several days, instead of spending the time choosing and acquiring more premade food each day? I’ve seen some folks online who have great success using catering as a meal prep option because it’s more frugal than ordering separate meals, but it also incurs less time investment as well as costing less money.

nim 12 Mar 2023 18:48 UTC
1 point
0
in reply to: sudo -i’s comment on: Minor Life Optimization: Consider Ordering Your Food To-Go
IMO the most obvious harm reduction strategy for “fast food delivery is expensive and terrible” is not to order different fast food, but to keep pre-made frozen meals on hand. You can buy frozen meals with the nutrition profile of your choice, make them yourself, or pay someone to make them for you. This costs less money and time than ordering delivery, and has the added benefit of leveraging that cognitive bias where you make “healthier” food choices when planning meals in advance compared to decisions that you make while hungry.

I’d postulate that people often order delivery because it’s the quickest and easiest option available to them. It seems like getting people (including oneself) to eat something healthier than their defaults is a matter of making something even quicker and easier available, rather than offering a choice between “do it your usual way” and a higher-friction option of checking a list first.

nim 12 Mar 2023 3:14 UTC
2 points
0
in reply to: sudo -i’s comment on: Minor Life Optimization: Consider Ordering Your Food To-Go
My experience is that not all locations (work etc) have refrigerator space conveniently available, but if you have access, that’s great!

I find that asking the person at the counter to give me a box is a much quicker operation than asking wait staff to put my food into a box for me.

nim 12 Mar 2023 2:20 UTC
11 points
5
on: Minor Life Optimization: Consider Ordering Your Food To-Go
If you plan to eat at the restaurant, you can just ask them for a box if you have food left over.

Also, ordering to-go and then eating there only works at places where you can just grab a seat. In many places I’ve eaten, saying “for here” is how you get allocated a place to sit if you want to eat there.

If you want to save even more time, phone the restaurant and place your order before you begin traveling to it. That way they can prepare your food during your travel time.

A potential con is that most food needs to be refrigerated if you want to keep it safe to eat for several hours. Depending on what happens between mealtime and snacks “later”, you may need to handle additional cognitive load to store the food appropriately.

nim 8 Mar 2023 23:34 UTC
2 points
0
on: Computer Input Sucks—A Brain Dump
I notice that I have two different speeds of thought: “raw ideas” and “coherent sentences”.

Writing at the speed of raw ideas is an interesting proposition because I don’t think humans have ever actually been able to do that. Checking this assumption, I find https://en.wikipedia.org/wiki/Words_per_minute with the un-cited claim that most handwriting tops out around 20wpm.

To my knowledge, the fastest that people ever tried to write by hand was by specialists using shorthand, which surprisingly the same wiki page claims can get up to 350wpm in competitions. That’s a bit like looking at the olympics and saying “humans can run at 27mph”, but setting a bound still seems useful.

For capturing ideas while walking, have you considered using voice recording and/or text-to-speech?

Personally, I touch-type pretty fast because for a few years in middle school, the internet was my primary means of social contact. I more or less learned to type as fast as I wanted to talk. Today, people who hear me typing often express surprise when they discover I was really typing and not just pretending to, because they thought it was too fast; for reference, a free online test just clocked me at 111wpm. I mention this because I do not personally have the problem of thinking in coherent sentences faster than my typing speed. Conversely, I often have to pause to think through what I’d like to say next, because my typing has out-paced the upcoming-sentences buffer in my thoughts.

When brainstorming, I can think ill-formed ideas faster than I can type them, but I address this issue by capturing such ideas using incomplete notes. I find that the process of turning a thought into a coherent sentence helps clarify and define the thought, and that whole process goes slower than my typing speed.

Considering that many of the traditionally extolled benefits of “writing” seem to revolve around that reification of ideas, I don’t think I understand the problem space in the same way you do. Have you identified an input speed that you’d consider adequate? How does the hypothetical adequate speed compare to your speed of speech?

nim 8 Mar 2023 23:10 UTC
1 point
0
in reply to: gjm’s comment on: Article about abuse in LessWrong and rationalist communities in Bloomberg News
I am a pessimist who works from the assumption that humans are globally a bit terrible. Thus, I don’t consider the isolated data point of “humans in group x have been caught being terrible” to be particularly novel or useful.

Reporting that I would find useful would ultimately take the form “humans in group x trend toward differently terrible from humans in other groups”, whether that’s claiming that they’re worse, differently bad, or better.

Whenever someone claims that a given group is better than most of society, the obvious next question is “better at being excellent to each other, or better at covering it up when they aren’t?”.

The isolated data point of “people in power are accused of using that power to harm others” is like… yes, and? That’s kind of baseline for our species.

And as a potential victim, reporting on misconduct is only useful to me if it updates the way I take precautions against it, by pointing out that the misconduct in a given community is notably different from that in the world at large.

nim 7 Mar 2023 22:36 UTC
3 points
−2
on: Article about abuse in LessWrong and rationalist communities in Bloomberg News
I read the first half and kind of zoned out—I wish that the author had shown any examples of communities lacking such problems, to contrast EA against.

nim 17 Feb 2023 21:27 UTC
1 point
0
on: Follow-up Posting on Cyborg Psychologist
Thanks for the reminder to check my email for the results from clicking through it yesterday!

Perhaps I am inadequately similar to the people it was tested on, but I find it quite disappointing. The output reads like a poorly designed horoscope—it feels like it’s tuned to emit claims which are supposed to resonate with anyone, and intermittently fail to resonate with me due to my own eccentricities.

In taking the test, I answered quickly on the questions where an answer seemed blatantly obvious, and paused longer on those where I felt uncertain and had to think more deeply. Despite this spread of response times, my result began with “Hmm … It looks like you just clicked through the test quickly, perhaps because you wanted to get a quick overview or take a look at the questions. In any case, this merits the caveat that the results will likely not be very meaningful, so you may consider re-taking the test!”.

Borderline incoherent and often self-contradictory, the rambling results only resemble therapy to the extent that performance art also does. There’s certainly a lot of potential in getting people to analyze themselves with a little automated prodding, going back to Eliza, but this app doesn’t give me the impression that it adds anything which couldn’t have been implemented pre-GPT.

nim 17 Feb 2023 20:58 UTC
7 points
2
on: Should we be kind and polite to emerging AIs?
I model my self-perception as being heavily influenced by the behaviors that I observe myself displaying.

I notice that I feel better about myself when I see myself treating others with consideration and respect.

When I work with animals, the tone of my speech to them is for their benefit, but the particular words I use are more to moderate my own mood and behavior than from any expectation that they understand the language. Similarly, treating possessions well and being relatively careful not to damage them is beneficial to oneself, because it saves the trouble and expense of replacing or repairing a needlessly damaged item.

On the other hand, there are situations where I think that exchanging pleasantries may make the AI’s job harder, kind of like how chatting socially with a human in certain situations can actually be impolite (such as if you’re holding up a line or distracting them from their work). I have never formed the habit of saying please and thank you to art AIs, because I have the impression that every token in their input contributes to the image output, so adding pleasantries that aren’t part of the request feels rude due to being distracting.

nim 12 Feb 2023 4:34 UTC
2 points
0
in reply to: the gears to ascension’s comment on: In Defense of Chatbot Romance
The specific failure mode I’m hearing you point a highly succinct reference toward is shaped like enabling already-isolated people to ideologically/morally move further away from society’s norms in ways that human interaction normally wouldn’t.

That’s “normally” wouldn’t because for almost any such extreme, a special interest group can form its own echo chamber even among humans. Echo chambers that dehumanize all humans seem rare—in groups of humans, there’s almost always an exception clause to exempt members of the group from the dehumanization. It brings a whole new angle to Bender’s line from Futurama—the perfect “kill all humans” meme could only be carried by a non-human. Any “kill all humans [immediately]” meme carried by a living human has to have some flaw—maybe a flaw in how it identifies what constitutes human in order to exempt its carrier, maybe some flexibility in its definition of “kill”, maybe some stretch to how it defines the implied “immediately”.

It sounds like perhaps you’re alluding to having information that lets you imagine a high likelihood of LLMs exploiting a similar psychological bug to what cults do, perhaps a bug that humans can’t exploit as effectively as non-humans due to some quirk of how it works. If such a psychological zero-day exists, we would have relatively poor resistance to it due to this being our collective first direct exposure to this powerful of a non-human agent. Science fiction and imagination have offered some indirect exposure to similar agents, but those are necessarily limited by what we can each imagine.

Is this in the neighborhood of what you have in mind? Trying to dereference what you’d mean by “validating dehumanizing actual humans” feels like being handed a note card of all the formulas for the final exam on the first day of a class that exists to teach one how to use those formulas.

nim 12 Feb 2023 4:22 UTC
2 points
1
in reply to: the gears to ascension’s comment on: In Defense of Chatbot Romance
Thank you for clarifying!

The editor didn’t spoiler your spoiler properly, if you were trying for spoiler formatting. I think some parts of society were kind of already, pre-AI, thinking in pretty great depth about the extent to which it can be possible to morally fantasize about acts which would be immortal to realize. “some parts”, because other parts handle that type of challenge by tabooing the entire topic. Examples of extant fantasy topics where some taboo it entirely and others seek harm reduction mostly hinge around themes of involving a participant who/which deserves not to be violated but doesn’t or can’t consent… which brings it to my attention that an AI probably has a similar degree of “agency”, in that sense, as a child, animal, or developmentally delayed adult. In other words, where does current AI fit into the Harkness Test? Of course, the test itself implies an assumed pre-test to distinguish “creatures” from objects or items. If an LLM qualifies not as a creature but as an object which can be owned, we already have a pretty well established set of rules about what you can and can’t do with those, depending on whether you own them or someone else does.

I personally believe that an LLM should probably be treated with at least “creature” status because we experience it in the category of “creature”, and our self-observations of our own behavior seem to be a major contributing factor to our self-perceptions and the subsequent choices which we attribute to those labels or identities. This hypothesis wouldn’t actually be too hard to design an experiment to test, so someone has probably done it already, but I don’t feel like figuring out how to query the entire corpus of all publicly available research to find something shaped like what I’m looking for right now.

nim 11 Feb 2023 18:28 UTC
4 points
0
in reply to: the gears to ascension’s comment on: In Defense of Chatbot Romance
Telling people what they want to hear, kind of like how starting with validation works in good listening?

Addictiveness is also a concern when food, drugs, media, or a culty social group is the only good thing in someone’s life.

Addictiveness seems to be less of a concern when someone has a lot of things they find good—the mental image of “healthy” and “well-adjusted” is based on balance between deriving benefits from many sources.

This raises the question of how much good it’d do to add a cheap, accessible, and likely less-harmful Second Good Thing to the lives of people who currently have only one Good Thing and the resultant problematic addiction type behaviors.

nim 6 Feb 2023 1:12 UTC
3 points
0
in reply to: lc’s comment on: What fact that you know is true but most people aren’t ready to accept it?
I hope it’s all right to butt in here—I think the animals I eat have inner lives, and the ones I raise for food are less tortured than the ones who live on factory farms, and also less tortured than those who live without any human influence. I think that animals who live wild in nature are also “essentially tortured”—those which don’t freeze or get eaten in infancy die slowly and/or painfully to starvation or predation when their health eventually falters or they get unlucky.

I think the humans who supply the world with processed food have inner lives and are essentially tortured by their circumstances, also. I think the humans who produce the commercial foods I eat, at all stages of the supply chain, are quantifiably and significantly less happy due to participating in that supply chain than they would be if they didn’t feel that they “had to” do that work.

If I was to use “only eat foods which no creature suffered to create” as a heuristic to decide what to eat, I’d probably starve. I wouldn’t even be able to subsist on home-grown foods from my own garden, because there are often days when I don’t particularly want to water or harvest the garden, but I have to force myself to do so anyways if I want it to not die.

I agree with you on the principle that torturing animals less is better than torturing animals more, but I think that the argument of “something with an inner life was tortured to make it” does not sufficiently differentiate between factory meat and non-meat items produced by humans in unacceptable working conditions.

nim 4 Feb 2023 21:58 UTC
1 point
0
in reply to: Vladimir_Nesov’s comment on: Path-Dependence in ChatGPT’s Political Outputs
Your phrasing gets me thinking about the subtle differences between “having”, “expressing”, “forming”, “considering”, and other verbs that we use about opinions. I expect that there should be a line where person-ish entities can do some of these opinion verbs and non-person-ish entities cannot, but I find that line surprisingly difficult to articulate.
Haven’t we anthropomorphized organizations as having opinions, to some degree, for awhile now? I think that’s most analogous to the way I understand LLMs to possess opinions: aggregated from the thought and actions of many discrete contributors, without undergoing synthesis that we’d call conscious.
When I as an individual have an opinion, that opinion is built from a mix of the opinions of others and my own firsthand perceptions of the world. The opinion comes from inputs across different levels. When a corporation or LLM has an opinion, I think it’s reasonable to claim that such an opinion synthesized from a bunch of inputs that are on the same level? A corporation as an entity doesn’t have firsthand experiences, although the individuals who compose it have experiences of it, so it has a sort of secondhand experience of itself instead of a firsthand one. From how ChatGPT has talked when I’ve talked with it, I get the impression that it has a similarly secondhand self-experience.
This points out to me that many/most humans get a large part of their own self-image or self-perception secondhand from those around them, as well. In my experience, individuals with more of that often make much better and more predictable acquaintances than those with less.

nim 3 Feb 2023 19:37 UTC
5 points
6
on: What fact that you know is true but most people aren’t ready to accept it?
There are a few areas where learning more about a topic has caused me to update my own beliefs into views nuanced and unfashionable enough that I prefer not to disclose them in any setting where others might feel that I was attempting to persuade them to change their own.

One of these areas is the food supply chain. It’s fashionable to point out that things would be better if everyone cut out fossil fuels, or ate organic and local, or whatever, and stop there instead of following the suggestion to its conclusions and side effects. Actually, the carrying capacity of the planet is contingent on modern agriculture, including a lot of genetically modified plants and synthetic fertilizers. “Better” methods, as we currently know them, would feed fewer people from the same amount of land. Nobody seems to like explaining what they think should happen to the couple billion extra people who exist in our current world and wouldn’t in their ideal one.

Another area is modern medicine. All I’ll say about this is that the implied isomorphism between what we “can” do and what we “should” do does not stand up to much scrutiny. Look up the percentage of medical professionals who have do-not-resuscitate orders, compared to the general population, and have a think about what that might imply.

nim

[Question] Are the LLM “in­tel­li­gence” tests pub­li­cly available for hu­mans to take?

[Question] Are the LLM “intelligence” tests publicly available for humans to take?