I think the problem here is interpreting whether the non-existential objections are nearest risks or the maximal risk from the objector. Suppose our roommate wants to bring a gorilla to live with us. I object that it will eat all our bananas. You say "Eat our bananas! Who cares it's going to rip our arms off while we sleep!"
The key here is that my objections can be interpreted in two ways:
-
(maximal risk) If we get the banana grabbing figured out, I'm on board.
-
(nearest risk) The banana grabbing already meets the threshold for me to veto it, I don't even to weigh all the additional risks beyond that, which I would if the banana grabbing was solved, and would still veto.
In this hypothetical, your response, assumes I mean 1, when I might mean 2. Why talk about whether the gorilla might kill us in our sleep when we can align around the banana thing and get the same outcome of no-gorilla. This is especially helpful if our fourth roommate thinks the night-murder thing is ridiculous, but can be convinced of banana grabbing concerns.
Screaming harder about night-murder and dismissing banana-grabbing as trivial, actually hurts the case with the fence sitter, who's name is Allan and loves the beach.
It gets worse when I say to Allan in concession, look there's like a 90% chance of banana grabbing and a 50% chance, of night-murders too. Then you jump in and scoff at my 50% as too low and of trivializing the real danger. Now we are in an inside baseball debate that simultaneously makes Allan take both banana-grabbing and night-murder less seriously.
I am not arguing about the likelihood of AI outcomes. Yud seems hysterical to me. But what do I know either way.
I am making the (apparently controversial) claim that I like the idea of someone who has the future of their children as a weight on their decision making algorithm when pursuing transformative technologies.
These are stakes in different futures for different people. Elon Musk has a perspective about human longevity that I am sympathetic to. When multiple groups of people have different future visions, each person is going to align to the leaders who most share their own.
Suppose three tech-billionaires all find a genie (it can be an AI genie if you want) who will grant them one only vision of the future of AI and humanity.
The first wants the fruits of humanity to reach the stars and survive trillions of years. The genie says the way for this to happen is for AI to succeed humanity, which may be destroyed in this process. The first finds this acceptable, echoing "I believe it should be regarded as a privilege to be a stepping stone to higher things". He believes these AI beings are our descendants and the future belongs to them.
The second wants a transhumanist future of long-lividness and maybe techno-immortality. The genie says that for this to happen, human reproduction will have to be bottlenecked to prevent Malthusian destruction. Un-exalted humanity will be culled and may die out as they will be of little use to the exalted, and represent a threat to their resources. The second finds this acceptable since has no need for descendants, as he will occupy their place.
The third isn't opposed to AI space explorers or transhumanist improvements but mostly wants his children and their children and theirs after to have the option to live their life in traditionally biological ways in peace and prosperity. He wants them to be able to form human families and create new generations. The genie says that this is doable but may altogether prevent or delay the opportunity for AI and transhumanists.
So all three futures are not necessarily incompatible, but only one gets to be prioritized. You can call all three of theirs "stake" in the future (though the first much less so), but you can see that each primary purpose comes at the expense of certainty of the other two.
DaeschIndustries and Chrispratt, seem stupified and angry at the idea that I might endorse the third guy, at the expense of the other two because this isn't dEmOcRaTic. I have my values and want to see them survive. Democracy is not a terminal value. Usually democracy is a great compromise, but on an existential scale, it can break down if your real values have an existential bottleneck.
I think this splits too quickly into a discussion about principled views that I would be happy to have under separate cover. I'd rather revert to my only real point that, as a parent the concerns of other parents about their children is a force of commonality and a potential for alignment. I recognize that in Elon Musk to an extent, and I was surprised to see both Elizier express sentiment that at least the child of a loved one is top of mind for him.
I am of course, aware of the ways appeals to children can be an emotional camel's nose into the tent of control. But my perspective is to ask, why it works and whether that reason is not always wrong.
People with kids, and people with traditional families (neither Elon, nor Elizier have the latter) are going to weigh future planning differently than those without. Am I, someone invested in the survival of the traditional human family wrong to prefer the leaders and those with power over transformative technology to share my experiences and values?
Generally speaking I want leaders and decision makers and people with power to share my values (as does everyone everywhere all the time. Just because the liberal's values are liberalism, doesn't mean that their desire for leaders to prioritize liberalism isn't the same exact impulse). AI is not an exception to that, and might be, rather the most extreme case in my lifetime.
I don't trust Altman, I don't trust Yud, and I don't trust you for the exact same reason.
If you don't have children and want to become a transhumanist immortal being, you shouldn't trust me (hypothetically. In reality, I have no power or agency and wouldn't make enemies over something I can't control).
You cannot be bothered to obscure the self-serving, gratuitously unprincipled nature of your words.
Self-serving? of course! So are all of your positions. Look I like liberal democracy. but I like it because it serves the world well, myself and my family included. The point at which it doesn't I don't have to religiously hold libertarian values.
Unprincipled? Absolutely not. This is a bullshit attack. My principles are based on values you disagree with, My positions which extend from my principles may be extrapolating on faulty data or predictive ability, but they exist. My principles are primarily toward the flourishing of my children and the of the existing human race. I think people with kids also have some extra buy in there. People without kids who want to appeal to democratic ideals, then use that to gamble the future of those with kids are less allied to my worldview.
Now I also have some WEIRD lifestyle preservation impulse. Because I do not come from Russia like you or India like selfmade, I am less inclined to rock the boat of my 'good life'. However, it is my Christian belief that lets me know that this particular self-interest is not morally acceptable past a very limited point. If you told me I could push a button that would preserve my lifestyle but keep the third world in poverty, part of me might like to, but I would not. Is that self-interest somewhat laundered through the 'altruistic' interest of my children. Yeah, and admittedly it becomes dicier there. But your interest in democratic ideals is likewise laundering of your own self-interest as well.
You and ChrisPratt both took the "cheeky" line too literally. I do not actually advocate a policy where only people with children get a stake.
Much more seriously, I am noting that Elon Musk's perceptions and goals about humanity are more readily parsable and agreeable to my POV than a childless technologist. Elon Musk has expressed a lot of views about human concern that I, (perhaps wrongly!) recognize as informed by the worldview of a parent, and that is a comfort against the rhetoric I find coming from a lot of other people. I said in my post that it could even be a product of my own bias, extrapolating too far gets what I called a "cheeky" heuristic, not an actual governance suggestion.
That folks without kids are so immediately hostile to the idea that folks with kids want to put the interest of their kids forward, is one of the biggest redpills against the techno-liberal worldview. I used to find the common argument is such circles that "think of he children is an emotive backdoor to authoritarianism", until I had chidren to think of. That doesn't mean I am an infinite safteyist. But it means I can recognize and reciprocate when other leaders are clearly thinking of the children.
You're taking what was explicitly called out as a cheeky framing of what is more of a heuristic for why I trust Musk more than Altman, people with kids more than single people when talking about the future of civilization and asking me to generalize it into a principle. But sure let's play with it.
All three of your examples are Mad-libs fallacies, they are written the same way, but actually point at the opposite of my argument (if taken as a 'principle)
-
'you don't get to have a say on abortion unless you have a uterus'
-
'you don't get a say in gun control unless you own an AR-15'
-
'you don't get a say in our adventures overseas unless you serve(d) in the military.'
The more accurate analogy that fits with your examples is something like "You don't get a say on AI, unless you are working on AI" or own a LLM or something
But again, that is very far away from what I said. None of those examples are formulated to capture what I was talking about. They all angle at direct experience in the subject, with the partial exception of the abortion one, but that will quickly develop into an abortion debate.
Your examples are of agency in the policy based on exposure to the tools, while mine was agency based on effects of the outcomes. Again the abortion one only follows if you argue that the baby isn't a party with exposure.
but in theory, self-determination and broad involvement of the populace in moral questions seems to be a fundamental value of the western political tradition.
So this is the part that I disagree, and my first round on the Motte helped disabuse me of. AI risk is a good example of where this kind of libertarian ethic breaks down.
My "general principle" looks something like this, but it's really a heuristic not a principle
-
If you are farming the commons, appeals to axiomatic autonomy and unlimited self-determination are weak.
-
EVERYTHING you do is farming the commons, though unequally weighted.
-
The more something farms the commons, the more it should be determined by those who's commons are affected than by the farmer's desires.
-
Something about how, if you extend this to longtermism, you've gone to far.
I understand, and am generally inclined to accept this point. Mostly, it is only my tribe with children who I trust the most. However, tribes are concentric circles, and here, I don't think your objection applies if we are talking about existential crises.
I am sure that Elon Musk would do a lot of (indirect) harm to me to see his children survive, thrive. But if AI has the possibility to conquer / destroy the whole world or flip the whole Western economy into unpredictable pieces, then my entire point is that a parent saving their own child, will necessarily care about saving at least some of humanity. Certainly more so than a singleton who abstractly reasons that a successor race of AI reaching the stars actually extends our legacy best.
If you have a button that might blow up the whole world or give you riches, I believe the risk reward calculus changes if you also have children and moreover the way you approach the question is less alien to me.
ON the other hand in the case of just economic board flipping, I think it's not coincidence that the two most hyped transhumanists on the board are non-Westerner expats from their own countries, and at least one of whom has no children. I trust the future most to those with families invested in a Western, Catholic, American life and move outward from there.
I posted, but deleted this in response to a previous AI thread, but I think it actually aged better with Elon's signature to the letter yesterday and Yud's oped:
I am not a Musk fanboy, but I'll say this, Elon Musk very transparently cares about the survival of humanity as humanity, and it is deeply present down to a biological drive to reproduce his own genes. Musk openly worries about things like dropping birth rates, while also personally spotlighting his own rabbit-like reproductive efforts. Musk clearly is a guy who wants and expects his own genes to spread, last and thrive in future generations. This is a rising tides approach for humans Musk has also signaled clearly against unnatural life extensions.
“I certainly would like to maintain health for a longer period of time,” Musk told Insider. “But I am not afraid of dying. I think it would come as a relief.”
and
"Increasing quality of life for the aged is important, but increased lifespan, especially if cognitive impairment is not addressed, is not good for civilization."
Now, there is plenty, that I as a conservative, Christian, and Luddish would readily fault in Musk (e.g. his affairs and divorces). But from this perspective Musk certainly has large overlap with a traditionally "ordered" view of civilization and human flourishing.
Altman, on the other hand has no children, and as a gay man, never will have children inside of a traditional framework (yes I am aware many (all?) of Musks own children were IVF. I am no Musk fanboy).
I certainly hope this is just my bias showing, but I have greater fear for Altman types running the show than Musks because they are a few extra steps removed from stake in future civilization. We know that Musk wants to preserve humanity for his children and his grandchildren. Can we be sure that's anymore than an abstract good for Altman?
I'd rather put my faith in Musks own "selfish" genes at the cost of knowing most of my descendants will eventually be his too than in a bachelor, not driven by fecund sexual biology, doing cool tech.
Every child Musk pops out is more the tightly intermingled his genetic future is with the rest of humanity's.
In Yud's oped, which I frankly think contains a lot of hysteria, mixed among a few decent points, he says this:
On March 16, my partner sent me this email. (She later gave me permission to excerpt it here.)
“Nina lost a tooth! In the usual way that children do, not out of carelessness! Seeing GPT4 blow away those standardized tests on the same day that Nina hit a childhood milestone brought an emotional surge that swept me off my feet for a minute. It’s all going too fast. I worry that sharing this will heighten your own grief, but I’d rather be known to you than for each of us to suffer alone.”
When the insider conversation is about the grief of seeing your daughter lose her first tooth, and thinking she’s not going to get a chance to grow up, I believe we are past the point of playing political chess about a six-month moratorium.
I'm unclear whether this is Yud's bio-kid or a step kid, but the point ressonates with my perspective of Elon Musk. A few days ago SA indicated a similar thing about a hypothetical kid(?)
I once thought about naming my daughter Saffron in its honor. Saffron Siskind the San Franciscan, they would call her. “What a lovely girl in a normal organic body who is destined to live to an age greater than six”, the people would say.
In either case, I don't know about AI x-risk. I am much more worried about 2cimerafa's economic collapse risk. But in both scenarios I am increasingly of a perspective that I'll cheekily describe as "You shouldn't get to have a decision on AI development unless you have young children". You don't have enough stake.
I have growing distrust of those of you without bio-children eager or indifferent to building a successor race or exhaulting yourself through immortal transhumanist fancies.
That's the charitable version (that she believes that porn is in fact empirically a bad thing, but she doesn't have enough data to prove it).
Maybe I'm confused, because i thought that's what OP was saying she was saying explicitly.
On a tangent, were I to sit down with someone like Aella and try to discuss whether porn was empirically bad, we'd never even get to the empirical part because we'd have such different worldviews that we wouldn't be able to operationalized 'bad'. I'm all for finding common ground, but most of the ground to cover on porn is too tied to conflicting foundational moral visions, that it's...well... about as effective as masturbating.
I think my example was too detailed, and the analogy gets lost. The TLDR is that Russell's teapot and Sagan's dragon are ontologically different concepts. And you can have anecdotal data of the former which can be used to diagnose lack of formal observation, but you cannot have it in the latter.
I am not trying to equivocate qualitative and quantitative research. But qualitative research can observe phenomena that that existing quantitative research may be unable to effectively construct, generalize, or have enough power to measure.
Your default to no, he loses, circumvents my entire point.
If your belief is Russell’s teapot, you’ve already lost. “At least it’s not Sagan’s dragon”. How would we know, with current tech they look exactly the same. You are multiplying entities beyond necessity.
You're missing my distinction. The classic form of Russell's teapot is that it is completely unobserved, but not materially unobservable, while Sagan's dragon is both. They only both look the same when they are both speculative. Russel's teapot can be qualitatively or anecdotally observed while Sagan's dragon cannot. If Megan's argument is that she's never actually seen the damage of prostitution, but is convicted it exists based on her ethical assumptions, then yes, it matters not whether it's a teapot or a dragon.
But if her conviction is based on cases, then it's more like, yes I have seen dishware in outer space, even can't give you the coordinates, so I'm not going to take your inability to find it with current satellite tech as proof it isn't there. except the analogy fully breaks down here so this last paragraph is more confusing than enlightening, so why am I even still typing, I'm not even using periods anymore, just commas,
I am going to wade back into the motte for a sec and respond to part of this. Bear with me, as my tie in will take a few detours. TLDR, I grow weary of the cult of “data-driven decision making”
There’s a difference between non-falsifiable theories , an non-demonstrable theories (unlikely the right term, I’m sure some rat has a real term for this).
The Sagan’s dragon is non-falsifiable, but the Russell’s teapot, even though it’s considered exemplar of unfalsifiable, is only non-demonstrable. It could be falsified, but we don’t have the tools to do it. People play these two interchangeably (often they can be), but too much and it causes a lot of soldier arguments. I think most of Caplan, who is correct a lot about education, a lot of his arguments play on a motte-and-bailey between these two.
Suppose Jon argued hard that learning Shakespear in middle-school paid off in various interpersonal interactions later in life. That is certainly not non-falsifiable, but it is almost a teapot's difficulty to measure empirically. Any study bound by real world constraints that attempted it would be insufficient.
So you say, it’s non-falsifiable, and Jon says No, I don’t think so. Jon goes out and interview a lot of people and puts together a nice phenomenology or narrative or whatever, and finds lots of anecdotal and circumstantial evidence of a phenomenon that appears again and again that many people seem to be able to draw a connection between their Shakespear and communication benefits. Suppose it is gold-standard qualitative research. Now I still think that it’s perfectly valid for you to stop here and argue, it’s not compelling enough to convince you.
But say you respond by pointing to several studies that went looking for these benefits but weren’t able to reject the null hypothesis of no connection. The first looked at learning Shakespeare and life outcomes with no relationship. Jon responds that of course the effect of a single course in shakespeare on life outcomes is going to be tiny, all other influences considered, that no study would be powered enough to find that signal. You find another study that looks at learning shakespeare and recall of his plays in college students, and finds very small retention. Jon again disagrees that it’s looking at the same thing. And so on.
You accuse Jon of refusing to update on data, and of holding a non-falsifiable belief. Here Jon admits that the whole logic model and all the influencing factors are somewhat unknown, but that there is connection as seen in his field research. Jon argues back at you that studies that don’t show any connection may be evidence that they aren’t designed properly since the phenomenon does exist and seems to in a nontrivial amount. He argues that if your data is correctly measuring the construct, it would predict that he wouldn’t have found the qualitative results he has.
He concludes that even with the unknowns, the benefit-cost is worth including it in the curriculum.
I’m not suggesting that Jon’s logic is air-tight, but I think it does show cracks in worshiping empirical ‘data’ in complex, longitudinal experimental problems, and the weakness of dismissing theories about difficult problems as unfalsifiable.
I think when someone like Megan says they won’t update on data, they’re essentially saying this. She has observed an actual and significant (not statistically) phenomenon that influences her epistemic and ethical view of the situation, enough so that when data that fails to capture it, her priors don’t rule out under-powered or poorly operationalized designs that aren’t measuring the right thing.
Another example. You ask me what data would change my mind that there are thousands or more faithful Catholics in the world, I would say none. Because I know several dozen myself. The alternatives that I live in a completely anomalous space and happen to know a large percentage of all faithful catholics, or that I am so bad at modeling others, the people I think are faithful aren’t, are both so ridiculously improba
- Prev
- Next

Sure but suppose you have 10 shows, and you want to add token diversity by having 10% LGTB representation.
You could make 1 of them have a central LGBT focus, you could make all of them have 10% lgbt themes or anywhere in between.
When you go with 'everything has to be a tentpole, you end up with worst of both worlds where 3 of them are LGBT focused and the other 7 have 10-20% LGBT themes, and you've lost any diversity that includes "not about LGBT"
When you decide that "Not-LGBT" doesn't have any place in your definition of positive inclusivity, fine, that's a value you can have, but you are clearly leaving profit on the table for the sake of values. There is clearly a large an audience in America that is interested in content which doesn't feature progressive values. If you go with the tolerance includes intolerance of intolerance view, fine. But stop the pretense of the 'profit-maximizer' explanation
More options
Context Copy link