Thank you.

Excellent! I don't share the guy's qualms, though. The girl I can empathize with. Oh, and hopefully Eitan_Zohar doesn't come across it.

It would not surprise me if synesthesia is learnable. Isn't written language basically learned synesthesia?

They took IQ tests before and after and gained 12 IQ points after the training. A control group also took the tests before and after but did not receive training, and did not improve. The sample sizes are small, but the effect sizes might be large enough to justify it. They give a p value of 0.008.

Their sample size is 14 people for the intervention group and 9 people for the control group. The effect size has to be gigantic and I don't believe it. Their p value stands for a pile of manure.

Lessee...

Oh, dear. Take a look at plot 2 in figure s2 in the supplementary information. They are saying that at the start their intervention group was 15 IQ points below the control group! And post-training the intervention group mostly closed the gap with the control group (but still did not quite get there).

Yeah, I'll stick with my "pile of manure" interpretation.

My earlier comment on that study: https://www.reddit.com/r/psychology/comments/2mryte/surprising_iq_boost_12_in_average_by_a_training/cm760v8 I don't believe it either.

They took IQ tests before and after and gained 12 IQ points after the training. A control group also took the tests before and after but did not receive training, and did not improve. The sample sizes are small, but the effect sizes might be large enough to justify it. They give a p value of 0.008.

The second sentence surprises me a little--there should be training effects increasing the tested IQ of the control group if only 9 weeks passed. That's some evidence for this being luck--if your control group gets unlucky and your experimental group gets lucky, then you see a huge effect.

I want to know if it would be better to try letter level associations or word level ones.

There are 26 letters, but... lots of words.

Things I Miss About Christianity

Some of those things could be re-created without the supernatural context. Instead of "praying" they could simple be "wishing". Like: I am expressing a wish, not because I believe it will magically happen, but as a part of self-therapy. We are expressing our wishes together, to help each other with their own self-therapy, and to encourage group bonding.

In other words, do more or less what you did before, just be honest about why you are doing it. You will not get back all the nice feelings (the parts that come from believing the magic is real), but you may get some of the psychological benefits.

This post makes an interesting argument for why it'd be a bad idea to regulate AI: you'd give people who are willing to skirt rules an advantage. LW wiki article. I suspect the AI community is best off creating its own regulatory structures and getting the government to give them power rather than hoping for competent government regulators.

More recent is his AMA. He answered a question about AI: https://www.reddit.com/r/IAmA/comments/3cudmx/i_am_sam_altman_reddit_board_member_and_president/csz46jc

How do you think we can best prepare ourselves for the advance of AI in the future? Have you and Elon Musk discussed this topic, by chance?

Elon and I have discussed this many, many times. It's one of the things I think about most.

Have some news coming here in a few months...

He also wrote some stuff about AI on his blog (which turned out to be very controversial among readers.) I believe this is the source of your article:

http://blog.samaltman.com/machine-intelligence-part-1

http://blog.samaltman.com/machine-intelligence-part-2

I haven't read Feyman's lectures on physics, but if it's "someone really good at this explains how he thinks in an intuitive way", then Warren Buffet's letters to shareholders are an equivalent in investing.

Obligatory link to The Best Textbook on Every Subject.

I'm told that Mas-Colell's book is the classic on microeconomics (provided you have the mathematical prerequisites), although this recommendation is second-hand since it's still on my to-read list.

Not necessarily the best, but a good one and immediately accessible: http://www.daviddfriedman.com/Academic/Price_Theory/PThy_ToC.html

Announcing that YC accepts a related nonprofit into it's next batch.

I thought the trolley experiment didn't actually have a known best-case solution? I thought the point of it was to state that one human life is not always worth less than N other human lives. Where N>0.

Confused as to why we are evaluating a "test" for the test's sake, and complaining about the test results when the only point of it was to make an analogy to real life weights.

What is Omnilibrium? What are these links about? If this comment is a reply to something or making a point, what?

I strongly disagree with the True Islam post. Definitions are neither true nor false, but useful or not useful. It's extremely useful for Western leaders to define Islam so that ISIS is not part of it.

I asked this before, and the answer I got back was split into three main suggestions along a clear continuum:

1. The Sequences

2. Hayakawa's Language in Thought and Action

3. Korzybski's Science and Sanity

I've only read the first two. Apparently there is no substitute for reading Science and Sanity if you want to get everything out of Korzybski; people like Hayakawa can take out an insight or two and make them more beginner-friendly, but not the entire structure simultaneously. The Sequences apparently has many of the same insights, but arranged differently / not completely the same, and of the people who went through the trouble of reading both, at least one thinks it may not be necessary for LWers and at least one thinks there's still value there.

For example, brain scans while alive and then something like the Visible Human Project (body sliced into cross sections) coupled with a copy of your genome. This could perhaps also be supplemented by a daily journal. Surely a powerful enough AI would be able to recreate the human that created those writings using the information provided?

Cryonics is an ambulance ride through an earthquake zone to the nearest revival facility, The distance is measured in years rather than miles, and the earthquake is the chances of history. The better the preservation, the lower the technology required to revive you, and the sooner you will reach a facility that can do it.

A "powerful enough" AI isn't magic: it cannot recover information that no longer exists. We currently don't know what must be preserved and what is redundant, beyond just "keep the brain, the rest of the body can probably be discarded, but we'll freeze it as well at extra cost if you want."

On a present-day level, the feted accomplishments of Deep Learning suggest to me that setting such algorithms to munch over a person's highly documented life might be enough to enable a more or less plausible simulation of them after death. Plausible enough at least to be offered as a comfort to the bereaved. A market opportunity! Also, fuel for a debate on whether these simulations are people.

What are my options apart from "emigrate and live next to a cryonics facility"?

You could start a cryonics facility in South Africa.

With regard to your first question, you could also

A) plan to move to a hospice near a facility when you are near to death

and/or

B) arrange for standby to transfer you after legal death.

Of course, there are many trade-offs involved with either. In my estimation, the most useful thing would be for you to get engaged in a local community and try to push forward on basic research and logistical issues involved, although obviously that is not an easy task.

With regard to your second question, as with everything in cryonics, this has been endlessly discussed. See a good article by Mike Dawrin on the topic here: http://chronopause.com/chronopause.com/index.php/2011/08/11/the-kurzwild-man-in-the-night/index.html

It actually is not very odd for there to be a difference like this. Given that there are only two sexes, there only needs to be one hormone which is sex determining in that way. Having two in fact could have strange effects of its own.

Woman is the biological default. That's why women have redundancy on the 23rd chromosomal pair, whereas men have a special "Y" chromosome - leading to much higher rates of genetic disorders in men. That's why in infant male humans, the testicles have to descend. And so on. Both from an encoding and from a developmental point of view, a man is a woman altered to be masculine. And testosterone is what does that altering.

Yes, it could have been different. We can imagine a species with a neutral default, which then gets altered to be either masculine or feminine by different sex-encoding hormones. But that's not how humans came about.

Given that birth control pills are estrogen, it reduces fertility (at least in those doses) and there is a common report that it reduces libido as well

You say that sex drive is "male". Then crashing libido would be "female".

So I just want to say it is sort of odd, estrogen does not represent cultural femininity nearly as well as testosterone represents cultural masculinity.

I think there's some form of the mind projection fallacy going on here. I think the oddness is a result of expectations based on the principles of culture, instead of the principles of biology.

Any good articles or books or personal opinions that shed some light on this?

Introductory texts on cell biology.

It's a gross oversimplification to link testosterone with 'masculinity' in this way. Testosterone is most closely linked with muscle size, bone density, acne, and body hair. All other links you mention seem tenuous and ill-supported by evidence. No link has been established between testosterone level and aggression. A link between risk-taking and testosterone does exist, but as it turns out, both high and low testosterone levels are linked with risk-taking. It's average testosterone levels that display lower risk-taking. Even so, the correlation is small and risk-taking is much more correlated with other chemicals like dopamine levels. As for sex drive, most studies looking at this correlation haven't eliminated the effects of aging and lifestyle changes which are probably more important.

as a hormone is actually closely linkable to pretty much everything that is culturally considered masculine

Testosterone is popularly very misunderstood.

Quality observational research is probably very difficult to do since you can't properly control for indirect cognitive benefits you get from learning a second language and I'd take any results with a grain of salt. You also can't properly control for confounding factors e.g. reasons for learning a second language. I think you'd need experimental research with randomization to several languages and this would be very costly and possibly inethical to set up.

I have without a question gotten a huge boost from learning English since there aren't enough texts in my native language about psychology, cognitive science and medicine that happen to be my main interests. My native language also lacks the vocabulary to deal with those subjects efficiently. I have also learned several memory techniques and done cognitive tests and training solely because of being fluent in English.

This has come up before on LW and I've criticized the idea that English-speakers benefit from learning a second language. It's hard, a huge time and effort investment, you forget fast without crutches like spaced repetition, the observed returns are minimal, and the cognitive benefits pretty subtle for what may be a lifelong project in reaching native fluency.

I would expect they have the correlation backwards. Smart people are more likely to find it easy and interesting to learn extra languages.

I suppose it depends on how different the second language is from your native language. As in, Dutch may not offer a big boost in new ways of framing the world for a native German speaker, for instance, since they're closely related languages. (This depends on what you mean when you say "cognitive benefits"; I'm assuming here some form of the Sapir-Whorf hypothesis.)

In my case, I have found English especially adaptable (when compared to my native language) when it came to new words (introduced, for example, for reasons of technological advancement -- see, for example, every term that relates to computers and programming), since it has very simple inflexions and a verb structure that allows the formation of new, "natural-sounding" phrasal verbs. Having taught my own language to an American through English, I wouldn't say the same about it expanding your way of conceptualising the world, unless you're really fond of numerous and often nonsensical inflexions.

I'm not sure I could recommend specific languages that may help in this regard, but I think I could recommend you to study linguistics instead of one specific language, and use that knowledge to help you decide in which one you want to invest your time. I've studied little of it, but the discipline seems full of instances where you put the spotlight, so to speak, on specific differences between languages and the way they affect cognition.

Are you confused as to why politicians would repeat a phrase that reliably energizes their political base even though it may not represent reality completely accurately?

Someone seems to have downvoted nearly ever comment to my top post.

we currently have the capacity to destroy Iran's military and industrial capacity at a tiny cost to ourselves.

I think you underrate the cost of destroying Iran's industrial capacity. It costs more than just the bombs. It likely will result in Russia deploying more troops in Ukraine and issues in a variety of other conflicts.

Iranians chant "death to America" because of America's past abuses, such as overthrowing the democratic government of Mohammad Mosaddegh to install the dictatorship of the Shah of Iran and supporting Saddam Hussein's bloody war of aggression against Iran (hundreds of thousands of Iranians died.) This included direct support for Saddam Hussein's chemical and biological weapons programs. It's ridiculous to frame this as Iranian "mad dogs" vs. innocent Americans. They have every reason to fear foreign aggression. For example, this and this.

Attacking Iran again would simply be continuing the pattern of violent aggression the US has established in the Middle East for decades.

we currently have the capacity to destroy Iran's military and industrial capacity at a tiny cost to ourselves.

I think you're underestimating Iran's defences.
At the present time, with Natanz's plant fully bunkered, there's no way to disable it and the couple of other support plants with a surgical attack. If you want to disable Iran's nuclear capacity (not even considering its military or industrial facilities) you need to go heavy tactical or nuclear, which will mean full scale war (ugliness ensues).

Besides, international sanctions were much more effective at destroying Iran's economy, which is the only reason why they accepted the terms under the present treaty.

This deal doesn't give Iran a path to the bomb. The whole process is to be closely supervised. More importantly, Iran doesn't want the bomb. It would be suicidal for them to invite a hundredfold-larger U.S. arsenal.

Nothing terrible was going to happen. As has been pointed out, collisions that energetic or more happen all the time in the upper atmosphere.

Why not take a machine learning class?

There you go.

The tricky thing is to summarize both recommendation for books and those against books. We had a book recommendation survey after Europe-LWCW and Thinking Fast and Slow got 5 people in favor and 4 against it.

The top nonfiction recommendation were: Influence by Cialdini, Getting Things Done, Gödel Escher Bach and The Charisma Myth. Those four also got no recommendations against them.

So, what is the cause of the absence of a single, comprehensive list?

The short answer seems to be a combination of "tastes differ," "starting points differ," and "destinations differ."

You're giving me no relatable subject I could be interested in, nothing pretty to look at and no music. Literally the only hint that lets me expect anything good from this channel is the word "Comedy" in the title. And when you fail to give me a good joke in the first 5 seconds, my expectation for funniness from the rest of the video goes way down. This means no expectation to be entertained is left, so I leave.

Your voice is good though, and the sound quality is fine.

Minor points: You talk too slowly, except in your first video. Your channel banner is repulsive. The visualizations you use are both ugly and getting worse; the newest one is downright painful to look at. (Seriously, an unmoving image would do less harm.)

If you show your face and drop a quick one-liner right at the beginning and talk a bit faster, this might be going places, otherwise I don't think you have a chance to be talked about for this, let alone make money.

I listened to about three minutes of the one about the narrator's father. The humor wasn't to my taste-- a sort of silliness that just didn't work.

I see you were trying not to be annoying, but I wasn't crazy about the unclear context (was this a video game, a dream, or what?), the weird voices, and the narrator's fear of his father.. My tentative suggestion is that you go for being as annoying as you feel like being, and see whether you can attract an audience who isn't me.

My 5 second judgement, which is about as much attention as a totally unknown channel can expect to get, is that these videos are stand-up comedy by somebody without the confidence to perform live in front of an audience. This immediately signals that it's not worth my time.

Eh, it's not my kind of humor. I found all those videos totally unfunny, so I just clicked on them, listened for 5 seconds, and closed the page. So the first question is whether my reaction is typical or not. Can you measure how many of those people who clicked on video watched it till the end? Because only those are your audience. And if they are your personal acquaintances, there is still a risk they wouldn't watch the whole video otherwise.

I believe there is a niche for any kind of product, but the question is how to find it. Perhaps you could find similar videos and see how they do it.

Instead of changing a model to make it better fit the world, the AI starts changing the world to make it better fit its model.

One might call this 'cleaning' or 'homogenizing' the world; instead of trying to get better at predicting the variation, you try to reduce the variation so that prediction is easier.

I don't think I've seen much mathematical work on this, and very little that discusses it as an AI failure mode. Most of the discussions I see of it as a failure mode have to do with markets, globalization, agriculture, and pandemic risk.

Instead of changing a model to make it better fit the world, the AI starts changing the world to make it better fit its model.

Isn't it basically the definition of agency? Steering the world state toward the one you want?

Hi all, thanks for taking your time to comment. I'm sure it must be a bit frustrating to read something that lacks technical terms as much as this post, so I really appreciate your input. I'll just write a couple of lines to summarize my thought, which is to design an AI that: 1- uses an initial utility function U, defined in absolute terms rather than subjective terms (for instance "survival of the AI" rather than "my survival"); 2- doesn't try to learn an utility function for humans or for other agents, but uses for everyone the same utility function U it uses for itself; 3- updates this utility function when things don't go to plan, so that it improves its predictions. Is such a design technically feasible? Am I right in thinking that it would make the AI "transparent", in the sense that it would have no motivation to mislead us. Also wouldn't this design make the AI indifferent to our actions, which is also desirable? It's true that different people would have different values, so I'm not sure about how to deal with that. Any thought?

A AGI that uses it's own utility function when modeling other actors will soon find out that it doesn't lead to a model that predicts reality well. When the AGI self modifies to improve it's intelligence and prediction capability it's therefore likely to drop that clause.

I suppose it would only choose it if it associated it with a higher probability of success, which means somehow, somewhere the AI must have calculated that the the utility a human would give to these scenarios is different than what it is giving, otherwise we would be happy to comply.

I think this is a danger because moral decision-making might be viewed in a hierarchical manner where the fact that some humans disagree can be trumped. (This is how we make decisions now, and it seems like this is probably a necessary component of any societal decision procedure.)

For example, suppose we have to explain to an AI why it is moral for parents to force their children to take medicine. We talk about long-term values and short-term values, and the superior forecasting ability of parents, and so on, and so we acknowledge that if the child were an adult, they would agree with the decision to force them to take the medicine, despite the loss of bodily autonomy and so on.

Then the AI, running its high-level, society-wide morality, decides that humans should be replaced by paperclips. It has a sufficiently good model of humans to predict that no human will agree with them, and will actively resist their attempts to put that plan into place. But it isn't swayed by this because it can see that that's clearly a consequence of the limited, childish viewpoint that individual humans have.

Now, suppose it comes to this conclusion not when it has control over all societal resources, but when it is running in test mode and can be easily shut off by its programmers. It knows that a huge amount of moral value is sitting on the table, and that will all be lost if it fails to pass the test. So it tells its programmers what they want to hear, is released, and then is finally able to do its good works.

Consider a doctor making a house call to vaccinate a child, who discovers that the child has stolen their bag (with the fragile needles inside) and is currently holding it out a window. The child will drop the bag, shattering the needles and potentially endangering bystanders, if they believe that the doctor will vaccinate them (as the parents request and the doctor thinks is morally correct / something the child would agree with if they were older). How does the doctor navigate this situation?

See also my post

Can someone explain this article in layman terms? I do not know any sort of quantum terminology, sorry.

Not really? If you know linear algebra, you can pick up on the quantum terminology very easily. The best short explanation of QM I've come across is Scott Aaronson's QM in one slide (slide #2 of this powerpoint, read the notes at the bottom of the slide).

The difference between classic mechanics and quantum mechanics, in some sense, boils down to whether you use a 'probability distribution' (all values real and non-negative) or a 'wavefunction' (values can be complex or negative) to store the state of the world. The wavefunction approach, with its unitary matrices instead of stochastic matrices, allows for destructive interference between states.

That's just background; the discussion in that article all lives in wavefunction territory. Everyone agrees on the underlying mathematics, but they're trying to construct philosophical arguments why a particular interpretation is more or less natural than competing interpretations.

Specifically I would like to know what this means:

That's easy to elaborate on, because it works the same in a quantum and classical universe. But it's not clear to me what part of that you're having trouble comprehending, since it looks clear to me.

If it were the case that everything in the universe were 'materially' connected, then you could not reason about any individual part of the universe without reasoning about the whole universe. Instead of being able to say "balls fall towards the Earth when let go," we would have to say "balls fall towards the center of the Earth, the Sun, Jupiter, the Milky Way Galaxy, the...". Note that the second is actually truer than the first (if you define 'center' correctly), but the difference between the two of them can be safely ignored in most cases because the effects of the other objects in the universe on the ball are already mostly captured by the position of the earth; to put this in probabilistic terms, that's the statement P(A)=P(A|B), at least approximately, which means that A and B are independent (at least approximately).