Friday, October 27, 2017

If You Take Parfit Seriously, You Should Commit Yourself To Creating Superintelligence

Cross posted at Speculative Nonfiction and Cognition and Evolution.

Derek Parfit makes the argument that if utilitarianism as it is commonly understood is to be taken to its conclusion - the greatest good for the greatest number - that mathematically we should care not just about making individuals happy, but making more individuals, to be happy. If you can have a world of a billion people all just as happy as a world of a million people, then that that's a no brainer.

The problem is when you get to the math of it. The "repugnant conclusion" that if the total amount of happiness is what matters, then you should favor numbers over quality of life. That is, a world of a hundred billion people with lives just barely worth living is better than a world of a hundred people with great lives - because the great lives are probably not a billion times greater than those of the hundred billion in almost total misery.

The obvious objection is that you're talking about theoretical people when you talk about those hundred billion. The counterargument is that we do care about theoretical people - our descendants - and you might already make environmental decisions to preserve the environment for the happiness of your grandchildren; right now you avoid (hopefully) littering the street to avoid upsetting people you've never met and will probably never meet.

There are other objections of course; for instance, that experienced happiness in an individual is what matters; otherwise slave plantations could be (in fact, probably are) morally acceptable.

But following Parfit's repugnant conclusion to its end, if the total amount of utility is what matters, then increasing the amount of utility possible to be experienced also matters. That is to say, there is no reason to stop at considering theoretical people, but rather we should consider theoretical kinds of experience, and theoretical kinds of experiencers. And there is nothing in Parfit's thesis provincial to or chauvinistic about humans. (If there were, that might solve the problem, because you could say "the closer something is related to me, the more I should be concerned with its happiness" - me and my brother against my cousin, et cetera - which, at very close genetic distances, is in fact what most humans already do.)

Therefore, we should try to make a world of a hundred million bipolar (manic) people who can experience hedonic value far in excess of what most of us ever do (assuming we can keep them manic and not depressed.) Or, even better, created an artificial superintelligence capable of experiencing these states, and not devoting all our resources to creating as many copies of it as possible. But cast aside those constraints - if you believe it is possible for a self-modifying general artificial intelligence with consciousness (and pleasure) to exist, then by Parfit, the only moral act is to give up all your recreation and resources to live in misery and dedicate your life to the single-minded pursuit of getting us one second closer to the creation of this superintelligence. The total suffering and happiness of life on Earth up until the moment of the singularity would quickly shrink to a rounding error, compared to the higher states these replicating conscious superintelligences might experience. Therefore, if you are not already singlemindedly dedicating yourself to bringing such a superintelligence to life, you are forestalling seconds of these agents' pleasurable experiences (which far offset your own suffering and maybe those of all living things) and you are committing the most immoral act possible.

This problem is superficially similar to Roko's basilisk (in the sense of your actions being changed by knowledge of a possible superintelligence) but I think it should still be called Caton's basilisk.

As a result of these objections, I do not think we need to take the repugnant conclusion seriously, and I do not think not dedicating yourself to creating a super-hedonic superintelligence is immoral.

Monday, October 23, 2017

The People Afraid of Their IQs

[Cross-posted to Cognition and Evolution.]

There has been some gnawing of tongues on the topic of IQ. Here I'm not talking about its very existence, which isn't open to debate (it's very real - disagree? Then you, Barack Obama, and Donald Trump are equally smart. I thought so.) It's that some people - young people, mostly - seem to be scared to find out what their IQ is, because it might not be as good as they would like, and then their life would be ruined.

These people are assuming a high degree of determinism. Yes, IQ is important (go to the link above and follow the links.) But at the same time, Warren Buffet has said that if you could trade every IQ point above 120 for money, you should. The greatest chess player in history Magnus Carlsen has said that too-high intelligence has been a handicap to some of his predecessors, who got distracted by other endeavors. Jeff Bezos of Amazon studied undergraduate physics at Princeton, noted how much harder he had to work than some of his peers, and switched out. What are we to make of this?

Let's start off by assuming that IQ is completely determined by factors out of your control, and furthermore completely determines your future. That being the case, learning one's IQ has been likened to being handed an envelope with your date and manner of death inside.[1][2] Anxiety-provoking? Sure, just like getting the result of an important test back - which, by the way, is invariably the direction the discussion goes, and no, no one likes the day the envelope comes, but everyone would be very upset if one year the nice people at ETS decided to rip up all the tests and give everyone an average. And just like the day you get your test results back, the only possible rational answer to whether you would want to know your death date is YES, and when people actually have such a choice in real life, they almost invariably DO want to know it. Cryonics crowd aside, you already KNOW you're going to die. You even have a rough idea of about when it will happen, statistically. You don't know what from, but you can make some guesses. When someone receives a terminal diagnosis, now they DO know what they'll die from, and they have a much narrower statistical window on when. So put yourself in this position: your doctor has just told you that you have pancreatic cancer. What is your VERY NEXT THOUGHT? "How long do I have?" You'd be pretty angry, and justifiably so, if she told you "I typically don't tell people the life expectancy because they might not want to know that."

There are many other mostly genetically predetermined attributes that strongly affect our quality of life, but the difference is that the rest are all obvious and easy to compare, so we can't remain ignorant of them; e.g., height, physical attractiveness, and to some degree wealth. Readjusting expectations and recovering from narcissistic injuries is hard, but for people who don't want to know their IQ, what's causing them to suffer is much more likely to be their anxiety tolerance and ego strength (which you can improve), than it is their intelligence (which you basically can't.)[3] Jordan Peterson has made the interesting comment that people who are more intelligent than the median in their communities unsurprisingly tend to elevate this trait above all others, but (more insightfully) because of the nature of intelligence as something that we use to build a worldview, people become unidimensional about human value and take intelligence as an end in itself rather than one of many traits to be prized, along with, e.g. impulse control or emotional stability, which are certainly not the same thing even if they co-vary. This tendency leads to these people who define themselves as "intelligent" avoiding communities filled with people smarter than themselves, and in so doing, limiting their own progress in life and the amount they can contribute to the world at large. It also leads to the ridiculous convention of clubs for smart people (Mensa.) Is there a club for tall people? No. Well actually yes. But it's not called a club for tall people, it's called the NBA, and they DO something with their tallness. And by golly, come to think of it, there ARE clubs for smart people (physics departments, Google engineering, medical schools, law firms, consulting agencies, etc.)[4]

Two specific reflections on this.

1) While genetics is the single greatest determinant of IQ, it is not the be-all end-all deterministic measure that many fear it is. In particular, "smart" may correlate with, but does certainly not equal, happy, moral, or successful, and furthermore there is an endless list of capacities, tendencies and talents that may correlate somewhat, but certainly not entirely, that determine the chances of success in various paths in life. I suspect that Peterson's unidimensionals get most upset in the U.S., where we most resent anyone telling us that any characteristic outside of our control has any influence over our future, despite the obvious reality.[5] Think of the trigger points in these discussions: race, socioeconomic class, upbringing, genes, the wiring in your brain - somehow the second we turn 18 we should be able to leave all the behind and emerge as an un-caused cause, right? (I wonder how much of the hostility to the very concept of IQ stems from this kind of thinking.) Below I've reproduced a figure in the Vox article (which in turn is reproduced from elsewhere) but what's interesting is how broad some of the bars are. What a broad bar should tell you is that there are more people in that profession (more outliers in absolute terms), and/or that IQ isn't as important. Case in point, protective service workers - intelligence appears to be less of a determinant there than for some of the others. I would expect ability to remain calm and vigilant and tolerate distress is at least as important.



2) As an aside, what Buffet was really saying is that there's an inflection point for the marginal value of IQ points around 120. We should grant that an additional IQ point in 2017 is more likely to benefit you than it was in 1917 - the world is hopefully a truer meritocracy, and a lot more value created by industries requiring intelligence rather than brawn or bravado. But there is still more likely to be an inflection point for the IQ vs wealth graph (as opposed to IQ vs overall utility - see #1 above.)


To the extent that IQ is a proxy for achievement, happiness, and self-worth, then just go try to do the things you're worried about and you will get rapid feedback - i.e., Scott Aaronson's recommendation that you just go try to do physics. "But what if I fail?" Think about it this way. How many people have actually said, "You know, if only I had an IQ test, I would've known better, and I wouldn't have made this massive career commitment that I now can't retreat from without massive damage to my finances, freedom, etc." But - interestingly - people certainly do say things like "If only I'd known X about myself, I would've chosen a different career," where X is something about your utility function, like the value you place on time vs. money vs. freedom, security vs. opportunity, ability to get along with certain personality types, etc. The key is to learn about your own many dimensions and find your selective advantage. How sad is Jeff Bezos that he couldn't keep up with the physics students at Princeton?



[1]If a genie told me I was going to die on a certain date and time from a certain thing, I would enjoy making life difficult for fate. Shark attack in 2022? I will move to the Mojave desert and not come out. Yes, I know, something will happen with a great white shark being transported by air between aquariums, and the plane will blow up and the shark will fall on me or something, but I'm really really going to make the universe work for it. I would grant my wife exclusive rights to my story in advance.

[2]While knowing your death date would help you plan, it would mess up the possibility of insurance, assuming everyone gets such an envelope and knows that everyone else does.

[3]If you're still not convinced that you shouldn't be crushed by your potentially low IQ, then consider that many people smarter than you and me both believe that by the end of this century, machines of far greater intelligence than any human who ever lived will exist, and none of the at-that-point meaningless differences in processing power between us flatworms will matter anymore.

[4]In a forum discussion I once referred to academia, medical schools, etc. as "clubs for smart people" and was immediately told "NO. Those are clubs for people with IQs 120-130." Yikes! I guess this particular Einstein had just cured cancer AND made a killing in the stock market that morning so they had time to police the forums for claims like this.

[5]Social media is not helping this, and I'm not the first to think there's a connection between prevalence of social media and the rise in social anxiety in Millennials. But I propose a specific mechanism which accounts for it, which is the failure of insulation between status hierarchies, or between layers of the same hierarchy. For example, growing up in an outer suburb of a rural state that doesn't touch saltwater, until the early 2000s it was easy to be blissfully unaware of the pecking order of academic, corporate and governmental hierarchies; you went to State U., got a good education, and worked in a nearby town without any sense that people in the Big City or at Corporate HQ were better or smarter. That's no longer the case, and already by age 15 there's no hiding from this reality. (I believe this is the hardest thing about growing up now as opposed to when I grew up.) I also think this explains part of the virulence of the culture wars, as each group, particularly the social conservative bloc in middle America, is suddenly aware that there are people who despise them, and that they're at the bottom of, and being relentlessly judged by, this status hierarchy they've suddenly noticed and which is ruled by exactly the wrong kind of people. (Contexts with overlapping status hierarchies would therefore be expected to improve happiness at least in the short term, and status hierarchy monoculture to make it worse; e.g. North Korea; also, professional training programs where you're at the bottom of the totem pole, spend all your time with people at the program, and have no social contacts outside the program.)]

Wednesday, October 11, 2017

Endowment Effect as a Rational Strategy Part 2: Don't Be A Sucker

Many seemingly irrational decision-making biases can be rational in some settings which differ from typical experimental settings. For example, psychologists and economists scratch their heads that in single-round games, people will expend resources to punish wrongdoers, even when there is no net benefit for the punisher. But in multi-round games, expending resources to punish wrong-doers may actually provide a net benefit and act as a deterrent against future wrongdoing. The irrationality arises because some of these behaviors are rigid with respect to setting, and even in single-round games where we'll never see people again (e.g. someone who cut you off on the freeway) we continue to punish wrongdoers regardless of setting or expected net benefit.

The endowment effect is a bias where we place irrationally high value on things already in our possession, as opposed to the open market. ("That car over there is a clunker, he'll never get three grand for it. But mine? Same year, same condition? Five grand easy. Great car.") Curiously, a group of hunter-gatherers without access to markets (i.e., roads) do not show the endowment effect. Their close relatives who do have access to markets, DO show the endowment effect. I speculated that this could be because the endowment effect is in fact a defense against information asymmetry. Somebody who buys and sells cars for a living knows how much your car is worth more than you do, and they can screw you. Consequently the endowment effect gives you a cushion. If you've never been screwed in an open market before, you don't develop this defense. (So goes my theory anyway.)

Sure enough, Data Colada brings us a paper by Weaver and Frederick which strongly suggests it's trying to avoid a bad deal which is what motivates the endowment effect.