Posts
Comments
From a historical perspective this is an excellent treasure cache. Truly when you are the cutting edge of something ideas, relationships, personality, and economics all truly come together to drive history.
I would not be surprised if lurking in the background of my thought is Tyler Cowen. He's a huge influence on me. But I was thinking of specific examples. I don't know of a good general history of "humanizing".
What I had explicitly in mind was the historical development of automobile safety: seatbelts and airbags. There is a history of invention, innovation, deployment, and legal mandating that is long and varied for these.
How long did it take between the discovery of damaging chlorofluorocarbons and their demise? Or for asbestos and its abatement - how much does society pay for this process? What's the delta between climate change research and renewables investment?
Essentially, many an externality can be internalized once it is named and drawn attention to and the costs are realized.
Well said.
The casual policing of positive comments about Sam Altman is unnecessary. Is this Sam Altman sneer club? Grok the author's intent and choose your own example. SA is a polarizing figure, I get it. He can be a distraction to the point of an example, but in this case I thought it made sense.
It is something for authors to be on the lookout for though. Some examples invite "missing the point." Sam Altman is increasingly one example of a name that invites distracted thoughts other than the point intended.
Yes I'm assuming a locally run open weight model will be useful, but not ultimately not sufficient for very complex tasks. I hope that something of the sort I describe can and will exist before too much regulation and optimized monetization occurs.
And we don't have good social models of technology for really any technology, even retrospectively. So AI is certainly one we are not going to align with human flourishing in advance. When it comes to human flourishing the humanizing of technologies take a lot of time. Eventually we will get there, but it's a process that requires a lot of individual actors making choices and "feature requests" from the world, features that promote human flourishing.
At my school all the students have to deposit their cell phones into pouches at the office the beginning of the day.. cell phone strictness alone does not create interpersonal chatter. You just also spend a lot of time and work creating an actual culture of conversation. It's an explicitly taught skill!
Your data on the elasticity of behavior matches my experience well..
Oh weird! I didn't create this one.
You are correct! Let me fix that!
I for one am definitely worse off.
- I now have to read Scott on Substack instead of SSC.
- Scott doesn't write sweet things that could attract nasty flies anymore.
Very helpful comment!
Upon further reflection my example was ill-chosen. The flashlight clobbering example doesn't trigger double-effect considerations, because there is only one immediate effect of the action - an unconscious intruder.
I wanted to use a simple example that had two effects , but I think you rightly point out that claiming the home-owner didn't intend to make the guy unconscious is specious sophistry. His unconsciousness and the safety of the home are two aspects of the same effect. The unconscious intruder entails the safety of the home.
If we claim they are different effects, then we would have to imagine that can be separated, but the safety of the home in the example requires the incapacitation of the intruder. There is no significant difference between claiming "I intended to knock the intruder out" and "I intended to do actions that make my home safe."
But do notice that if we use the definition of intention from OP, which I will stick to, that our intention was to achieve a safe home by using force to incapacitate the intruder with the attitudes appropriate, no gleeful pleasure-taking in finally getting the chance to clobber someone. I think everything in your final paragraph is correct about attitude and steps taken minimize unnecessary harm.
Our intention is the goal and will (which includes attitudes) to perform the acts to achieve that goal.
If we go to the terrorist compound case, then there are multiple effects: the deaths of the civilians and the death of the terrorist. These are separable in that the death of the civilians does not cause the death of terrorists nor vice versa. The same missile launch causes both. Because they are separable effects, it is possible to intend/will one effect, without willing the other effect.
My original example of the home intruder failed to identify two separable effects, but only a logical entailment of the clobbering. In that case, the considerations are simple proportionality, not DDE. Obviously, if I knock someone unconscious, I can't then beat him to death. I have already achieved the goal. If I shoot him, he might die. But that may still be proportional. If I shoot him and then unload two magazine rounds into him, that would be disproportional.
What the SEP article means by "the good effect is not achieved through the bad effect" is not well-defined. But generally people in the lit interpret it to be the abstraction of "No capturing civilians to perform useful medical experiments." I might rephrase the abstraction to, "No doing an action that has bad effects, in order to do a separate action that has good effects." (But that's tentative).
Thanks for commenting. Welcome.
You right on both accounts.
There are some agreed upon universal principles that societies do develop. "No ****ing the young. No randomly killing people."
Categorical imperative is exactly what I what I had in mind. However, the CI is not based upon what's acceptable to the actor or to the society at large, but rather according to reason. "Act only on the maxim which you can at the same time will that it should be universal law."
Your action only makes your home safe by means of knocking him out. This is in fact a nice clean example of something that is not permitted by traditional formulations of the doctrine of double effect, at least not if knocking the intruder out is considered bad.
It is precisely this instance for which DDE was formulated: to explain why it is permitted! https://aquinas.cc/la/en/~ST.II-II.Q64.A7.SC
Flashlight clobbering produces the good effect and the bad effect. They are coterminous events, even though one is logically necessary for the other. As you say, "a safe home by means of an unconscious intruder."
You are also right. I shouldn't have used "Thomistic" in the first place. But because real discussion of this gets started in Thomas' work (his account is unclear and too curt), I introduced the term, and thereby introduced a lot more confusion than I intended!
Double-effect is a broad term, not every moral philosophy has discussion of double-effect. Xunzi's Confucianism, Kant's Groundwork, Aristotle's Nicomachean Ethics, Plato's Republic all of which I have read recently do not have such a discussion. Kant, I guess, kind of does, but for Kant consequences, allegedly, don't bear on the morality of an action at all.
Of course, you and I whom I presume both hold utilitarianism in at least somewhat high regard expect ethical systems to deal with these questions in detail.
[1] I am not exactly sure what you think I think is meant by intent. Intent means goal. It is debated whether the specific actions taken by the agent should be considered part of intent or not. In the OP, I said it did, especially for epistemic reasons. I argued that specific actions chosen reveal the intent and that is why we should care about "the act itself."
Am I making any headway here?
I would like to find our cruxes, if we in fact have any.
BTW: double-effect discussion is not an RCC thing. It is used in hospital system decision-making, reformulated by professors at Harvard School of Medicine, and used as a framework by military professionals and international law courts.
Like, despite our internet space having a lot antipathy towards these types of discussions. Crafting a better and clearer framework is a worthy project and one I wish to do, at least for my own benefit if not others.
Standard modifies Thomistic, it's not normative as in "this is the standard", it's descriptive, as in "this is how the philosophy books and articles on it generally work".
I use that as a starting point, because it is the one which has discussion of double-effect. I privilege "double-effect" as a concept because it is the term which describes thinking through actions which have both positive and negative consequences, and also takes account of common sense things people care about like intent, means, and circumstances. Unless a person is a moral purist, one needs some account of double effect. Greater good benthamite utilitarianism would be one such account. I find such an account insufficient for guiding or analyzing action, but I'm not a priori beholden to any particular interpretation of what should be contained within an account of double-effect.
I think you misunderstood what is meant by intent. It's not a statement one makes to oneself about what one is doing. That'd just be self-deception.
There are interesting questions about self-deception, though. As Bryan Caplan(?) recently wrote, "You never see politicians talk about 'acceptable rates of human tragedy.'" To what extent do we play-act morality in order to generate the feeling of "clean hands?" When is that necessary? When, if ever, is it beneficial?
What is the best way to build a liquid market of DACs?
For these to be and competitive a lot of people need to see them.
An excellent quote: "a market-based society that uses social pressure, rather than government, as the regulator, is not [the result of] some automatic market process: it's the result of human intention and coordinated action."
I think too often people work on things under the illusion that everything else just "takes care of itself."
Everything requires effort.
Efforts in the long-run are elastic.
Thus, everything is elastic.
Don't take for granted that industry will take care of itself, or art, or music, or AI safety, or basic reading, writing, and arithmetic skills. It's all effort - all the way down..
What has more elasticity - birth rates or institutional and educational quality of the sort that leads to innovation? Clearly both are fairly inelastic in the short run (the shorter the run, the closer to perfect inelasticity), but in the long run a lot might be possible along both axes.
Metascience should focus on the institutional and educational quality questions (including creating better idea discovery processes!). Insofar as metascience depends on fertility, it is a heavily intermediated causal pathway, both by time, by generational differences and culture, by immigration policy, and by educational opportunities. All of which probably have greater elasticity in the short run than fertility. So I disagree with the broad brush used here.
Immigration's effect on metascience is moderated by types of skilled immigration in the short run, not immigration as such. This is not a nitpick and not merely important practical political reasons. Conceptually, immigration is a big broad concept that should be broken down into its constituent mechanical parts, which includes admission of people based upon some quality, whether it be genius, family ties, lottery, bribe, or "points."
Natalist and Immigration public policy people should work on their issues without much regard for 3rd order or 4th order effects for metascience. I think keeping these separate is conceptually important, even if they do ultimately causally bear on one another. Everything correlates with everything else.
The best we might do on the public policy level is nudge toward subcultures that are both scientifically advanced, epistemologically healthy, technologically cutting edge, and reproducing themselves at some above replacement rate through both education and children. If we nudge each part of the causal chain a bit towards greater fertility, high skilled immigration, and better institutions it could entirely counteract the trends you describe.
We don't want to Canticle for Leibowitz and Children of Men ourselves to death. But I think the world model being proposed around fertility and technology by, say, Robin Hanson, is missing a good causal diagram that includes the elasticity of countervailing forces. I'm not saying the issue is not a big one, perhaps the biggest one. But I am terribly unimpressed with most thinking on this topic so far.
Since we are at no moment capable of seeing all that is inefficient and wasteful, and constantly discover new methods of wealth creation, we are at each moment liable to be accused of being horribly wasteful compared to our potential, no? There is no way to stand up against that accusation.
The implication: overhangs CANNNOT be measured in advance (this is like a punctuated equilibrium model); they are black swan events. Is that how you see it?
Forming a nonprofit is not that difficult. It's like four extra hours of work to get the 501c3 status and a decent time delay of several months. Having someone else to fill out the 990 for you is nice, though!
Ah, of course. That makes sense.
I know sincere intelligent Christians who would just be relieved and respect that you've actually thought about the question of deism, and see that as a positive sign of intelligence, maybe even truth seeking?
I stand by this comment.
What could cause me to change my mind? Here are my cruxes.
If character assessment posts about particular people can be shown to cause a useful actions or ways of thinking for readers more often than they distract readers by unverifiable gossip.
If character assessment posts about particular people is used as a case study for reasoning about particular people to teach a broader lesson.
If character assessment posts about particular people allows community members to protect themselves from a real danger.
However, my beliefs are that these types of posts are juicy gossip that fuel idle speculation and status hierarchy games and serve no purpose except to make those who engage with content worse people who think more simplistically about human behavior and motivation. Even though this particular post is done fairly well for what it is, I think it is "bad form" and, perhaps, on the wrong site.
When I saw the topic, my first thought is that the epistemics of discussions of this sort (he said - she said stories about sins and perceptions) are inherently bad and cause more harm to those who engage with them than good. But the post isn't terrible quality.
Nonetheless, I am pre-committing to downvoting any future post about the personal relationships of famous people, which I take to be the category of thing, I am objecting to.
In traditional charitable enterprises, we just call it "a donor gift."
"Moral imperatives" is not a category that relies upon financial means. Moral imperatives in traditional Kantian framework are supposed to be universal, no? Just because some action could be personally and socially very beneficial doesn't make it morally compulsory. The benefits would have to be weighed against opportunity cost, uncertainty, game theoretic considerations, and possible contrary moral systems being correct.
Source was in a meeting with her. The public record of the meeting transcript should be forthcoming.
I'll be selling Dominant Assurance Contracts Insurance. If a post is not up to standards then you can file a claim and receive compensation. Power will shift towards my insurance company's internal adjudication board. Eventually we'll cover certain writers but not others, acting as a seal of quality.
If they pay you with PayPal and you pay them back with PayPal, then there is a 6% loss. It would be easier and less dead weight lossy to use a private challenge mechanism within Manifold. Or even to just create a market in Manifold and bet it down super low, and then resolve it against yourself if you get 10 bettors. That's a first thought... One objection is that since assassination markets don't seem to obtain, maybe they wouldn't work with blog posts either. However, in this case the market maker has both the market incentive and some other motivation not entirely captured by the market to complete the blog post.
I learned those from Wittgenstein, Aristotle, and Umberto Eco.
Numerical obsession can simplify important semantic subtleties. And semantic pyrotechnics can make my reasoning more muddled. Both are needed. As is, I suppose, a follow up post on the non-LW ideas that pay a lot of rent.
One way to do this would be to create houses of study dedicated to these exams for students and a tutor work together in the community to accomplish these goals without requiring a very large costly institution. Group house plus the tutor/academic coordinator.
Some fields only require completing a series of test for entry. No degree required. I'll put in parentheses one's that I'm not sure don't require a bachelor's degree. Certified Actuary Chartered Financial Analyst (Certified Public Accountant) (Various other financial certifications) (Foreign Service Officer's exam) (The bar exam: I don't know how one can get them to let you sit the exam without a law degree, but it is allegedly possible in California, Vermont, Virginia, and Washington). There are a lot of certificate programs out there for long established work that involves brains but not in person learning (money and law). In computer science, "building things" is the certificate, I suppose.
I agree that if you are looking at it in terms of art generators that it is not a promising view. I was thinking hypothetically about AI enabled advancements in the energy creation and storage and in materials science and in, idk, say environmental control systems in hostile environments. If we had paradigm shifting advancements in these areas we may then spend time implementing and exploiting these world changing discoveries.
Maybe another perspective on point three is the additional supply of 2d written and 2d visual material will increase the price and status of 3d material, which would equilibrate as more people moved in to 3d endeavours.
So might this be a way to increase not only the status of atoms relative to bits, but use bits to reinvent the world of atoms through new physical developments? And if the physical developments are good enough and compounding would that stall the progress of AI development?
While I would say your timeline is generally too long, I think the steps are pretty good. This was a visceral read for me.
Some sociological points:
-
I think you don't give anti-AI development voices much credence and that's a mistake. Yes, there will be economic incentives, but social incentives can overcome those if the right people start to object to further specialized LLM development.
-
Although you have a fairly worked out thought on AI development, where the path is clear, for AIS the fact that you ended with a coin flip almost seems like slight of hand! The scenario you describe was clearly heading towards self-destruction, not just for narrative tenor and story based reasons, but for the lack of finding a foundation upon which to ground true AIS.
-
With improved robotics, there might be solutions to escape certain AI social traps by changing the balance of power between information and physical world development. I have in mind the idea that AI seems incredibly valuable today because physical progress has been so weak. But if physical progress becomes strong and fast, then allowing/forcing AI to slow down will be socially easier. It's an elasticity of demand model for competing goods. Does that make sense? If not, I'd happily elaborate.
Just read your latest post on your research program and attempt to circumvent social reward, then came here to get a sense at your hunt for a paradigm.
Here are some notes on Human in the Loop.
You say, "We feed our preferences in to an aggregator, the AI reads out the aggregator." One thing to notice is that this framing makes some assumptions that might be too specific. It's really hard, I know, to be general enough while still having content. But my ears pricked up at this one. Does it have to be an 'aggregator' maybe the best way of revealing preferences is not through an aggregator? Notice that I use the more generic 'reveal' as opposed to 'feed' because feed at least to me implies some methods of data discovery and not others. Also, I worry about what useful routes aggregation might fail to imply.
I hope this doesn't sound too stupid and semantic.
You also say, "This schema relies on a form of corrigibility." My first thought was actually that it implies human corrigibility, which I don't think is a settled question. Our difficulty having political preferences that are not self-contradictory, preferences that don't poll one way then vote another, makes me wonder about the problems of thinking about preferences over all worlds and preference aggregation as part of the difficulty of our own corrigibility. Combine that with the incorrigibility of the AI makes for a difficult solution space.
On emergent properties, I see no way to escape the "First we shape our spaces, then our spaces shape us" conundrum. Any capacity that is significantly useful will change its users from their previous set of preferences. Just as certain AI research might be distorted by social reward, so too can AI capabilities be a distorting reward. That's not necessarily bad, but it is an unpredictable dynamic, since value drift when dealing with previously unknown capabilities seems hard to stop (especially since intuitions will be weak to nonexistent).
Hey ya'll who RSVP'd, if you are interested send me a dm with your email and I will add you to our email list.
We are doing two events before the ACX Meetup Everywhere in October. You can learn about them by getting on the email list. This Thursday September 22 we are doing an online event in Gathertown 8pm eastern, 5pm Pacific. Feel free to send this out to people who would like this sort of thing.
https://app.gather.town/events/pXVcEMSts1dcxnOc7rqU
Loose Discussion Topic:
What does it mean to improve your local community/environment? Is local improvement possible? What would that mean? Should you care? Does local matter anymore? And how would one go about making such improvements possible? Come puzzle over these things and more with us Thursday night.
We tend to discuss the topic for a while and then move to whatever fits one's fancy. Also, gathertown allows splitting to talk in different groups. So that's, like, helpful if you want to talk Stable Diffusion, but another group of people wants to talk development economics, and another science politics.
Thank you for this high quality comment and all the pointers in it. I think these two framings are isomorphic, yes? You have nicely compressed it all into the one paragraph.
There's definitely evil AI.
I agree that thinking critically about the way AGI can get bottlenecked by physical processes speed. While this is an important area of study and thought, I don't see how "there could be this bottleneck though!" matters to the discussion. It's true. There likely is this bottleneck. How big or small it is requires some thought and study, but that thought and study presupposes you already have an account for why the bottleneck operates as a real bottleneck from the perspective of a plausibly existing AGI.
Bizarre coincidence. Or maybe not.
Last night I was having 'the conversation' with a close friend and also found that the idea of speed of action was essential for explaining around the requirement of having to give a specific 'story'. We are both former StarCraft players so discussing things in terms of an ideal version of AlphaStar proved illustrative. If you know StarCraft, the idea of an agent being able to maximize the damage given and received for every unit, minerals mined, and resources expended, the dancing, casting, building, expanding, replenishing to the utmost degree, reveals the impossibility of a human being able to win against such an agent.
We wound up quite hung up on two objections. 1) Well, people are suspicious of AIs already, and 2) just don't give these agents access to the material world. And although we came to agreement on the replies to these objections, by that point we are far enough down the inferential working memory that the argument doesn't strike a chord anymore.
We have a reservation for 8 at 1pm. I am wearing a blue tshirt, that says 'nihilist', a copy of Unsong, and an infant strap.
Come even if you feel nervous or shy. We will have fun, good conversation.
I have been reading and thinking about the ontology of chance, what makes a good introduction to chemistry, and the St. Louis County Charter.
These are good points! I have been thinking the same thing. However, I don't imagine the upper institute requiring prerequisites, just an entrance exam. But a four year college offers basically the same thing except they lower transaction costs to basically zero or making that decision to commit to something you like. Hence declaring or changing majors is usually easy if you do it sophomore year.
The price disclosure issue isn't a problem. You can Google average cost of any private college and it will give a good ballpark estimate which matches the OPs 20k+ chart. Colleges engage in near perfect price discrimination. It's not really considered nefarious, because it's both redistributive and expands supply. The richer pay more, thus subsidizing the poorer. This allow more students than otherwise would be able to to afford the college.
This price discrimination expands supply by increasing the absolute quantity of students who can afford the schooling there. Charging 20k for everyone would allow fewer students to attend than charging 30k for some and 10k for others.
That's true at the prestigious four year colleges. But there are hundreds of private four year colleges. Their supply of students is stagnant and beginning to backslide. If you talk to private four year college admissions officers, many are afraid of the coming great contraction in school aged people. Only Texas isn't having a contraction.
In any case, in John's model, that coming contraction should result in a decrease in number of specialized courses. We'll see. Courses might be somewhat sticky though.
[correction] "You are not allowed to teach that Purgatory is not part of Catholic teaching, because it is."
Martin: "Why should I not be allowed to teach it? I am allowed to debate it in the classroom."
"The disputations are one thing. Public tracts are another."
"You're just mad because I called you corrupt."
"Yes, that makes it easier to want to suppress you. Though we never officially censor people for saying that."
You seem to value loyalty and generally want to be a tit-for-tat player + forgiveness. That makes sense, as does calibrating for the external environment in the case of PDs. You also view the intellectual commons as being increasingly corrupted by prejudiced discrimination.
This makes sense to me. This makes sense, but I think the thinking here is getting overwhelmed by the emotional strength of politicization. In general, it is easy to exaggerate how much defection is actually going on, and so give oneself an excuse for nursing wounds and grudges longer than absolutely necessary. I'm guilty of this and so speak from some experience. So I think there's an emotional mistake here. Perhaps fixed by seeking out "philosophical repose", a bit more detachment from reacting to the opinions and thoughts of others.
Intellectually, I think you are making an important mistake. Cooperation in PDs and moral philosophy are two different, though related, things. Welfarist utilitarianism does not require impartiality, but choosing those actions which maximize good consequences. Since your reason for rejecting welfsrism is because of the consequences of not punishing free riders, you have not yet rejected utilitarianism. You are rejecting impartiality as a necessary tenant of utilitarianism, but that is not the same thing as rejecting utilitarianism. You have to go one step further and give non-consequentialist reasons for rejecting them.
Hope that helps.
I think the OP's overarching concern is something like a narrow utilitarianism whose decision algorithm takes EV over only a limited number of horizons and decision sizes. There is unknown EV in exploring the world more personally and in reproducing knowledge and skills. My hunch is that such optimization of human life takes these different aspects at least multiplicatively.
Expected value calculations have limits for decisions which will affect your worldview, i.e. exploration. Or decisions along the axis of goods which you don't have a good model for, i.e. education.
I am noticing some interdisciplinary additions perhaps on the history/sociology side:
Theories of cultural progress and the sociology of scientific/innovative community.
The education of excellent people.
The history of innovative communities.
I'm curious, Jason, what the best arguments you have found so far about the relationship between long-term trends in population growth and progress.
Updateless Decision Theory allows for acting as though you need to cooperate with an agent beyond you, even if it has a low probability of existing. I suppose your case of grandchildren works like this? I can cooperate with my as yet nonexistent grandchildren by making the probability of their existence higher, they will likely reward me more?
I'll have to work on my family norms then! Ancestor worship, it is!
Hi Mack,
You seem somewhat new. I just want to let you know that community standards of discourse avoid appeals to authority, especially appeals to authority without commentary. A comment like this provides little value, even if in jest.
Downvoted because just running in and dropping a scripture quote without commentary degrades LW conversational norms. This is not Wednesday night bible study and people don't nod their heads smilingly because you found a related scripture quote. Even if the audience were 90% believers, I doubt they would interpret scripture the same way you do. You should explain why you chose this quote and what bearing it has on turchin's admittedly glib point.
Besides switching from protestantism to at least something with a bit more harumph like, catholicism or orthodoxy, I encourage you to wrestle with the sequences, if you haven't already.
Regards!