Article Summary: Current and Near-Term AI as a Potential Existential Risk Factor

André Ferretti

LESSWRONG
LW

Comment Permalink

It's not possible to explain what you don't know, to answer a question you can't state, and "intelligence" doesn't save from this trouble, doesn't open the floodgates to arbitrary helpfulness, resolving any difficulties you have. It just does its thing really well, but it's up to its designers to choose the right thing as its optimization criterion. Doing the wrong thing very well, on the other hand, is in no one's interest. This is a brittle situation, where vagueness in understanding the goal leads to arbitrary and morally desolate outcomes.

See in context

16 What are you working on?

by jsalvatier

6th Oct 2011

1 min read

16

This is the fifth bimonthly What Are You Working On? thread. Previous threads are here. So here's the question:

What are you working on?

Here are some guidelines:

Focus on projects that you have recently made progress on, not projects that you're thinking about doing but haven't started, those are for a different thread.
Why this project and not others? Mention reasons why you're doing the project and/or why others should contribute to your project (if applicable).
Talk about your goals for the project.
Any kind of project is fair game: personal improvement, research project, art project, whatever.
Link to your work if it's linkable.

Personal Blog

16

What are you working on?

0Swimmer963 (Miranda Dixon-Luinenburg)

3Swimmer963 (Miranda Dixon-Luinenburg)

New Comment

97 comments, sorted by

top scoring

Click to highlight new comments since: Today at 6:30 PM

Some comments are truncated due to high volume. (⌘F to expand all)Change truncation settings

[-]jimmy14y160

I've been working on understanding and using hypnosis, which is better described as conditioning arbitrary associations through mind games.The theory and results have been quite promising, to understate it.

There has been a lot written on LessWrong about how one should think, but relatively little on how to actually implement these thinking patterns. One of the directions that I want to take this is helping people take these skills and actually implement them on the 5 second level. I also want to see if I can develop a viable 'rationality seed' by installing a gut level appreciation for the importance of taking ideas seriously (and in particular, the idea that rationality matters) as well as a few other select habits.

I've made a ton of progress and I feel like it's almost a new branch of engineering. I have concrete ideas to test and avenues to explore and not enough time to explore them all at once.

I'm looking for other people that would be interested in being brought up to speed on the theory and then running experiments with me. I'm excited to see where this can be taken :)

PM me if you are interested.

2jsalvatier14y

I would be very interested in hearing some specifics.

[-]spqr0a114y150

For over a year now, making full use of my body has been a big hobby. To this end, I am learning silly human tricks. It has taken a few weeks of sporadic effort and I am now able to snap my fingers consistantly. My next goal is to be able to whistle. I have made little progress so far and am not yet able to whistle particular notes. Previously successful projects of this type were refining my senses of smell and temperature. Now I can reliably tell the temperature of water or air to the degree C with a range of roughly -20 to 45C. Calibrating my sense of temperature was easy enough; I would feel something, guess the temperature, then read off a thermometer. If I was wrong, I might say something like "Oh so that's what 23 degrees feels like." and then feel it again.

Other ongoing projects of this sort are to smell relative humidity and to sneeze only in the presence of a bright light. I've lost a bit of progress since last thread, now darkening my field of vision works to prevent a sneeze, but I still sneeze occasionally at normal indoor room brightness. When I feel like sneezing I look at a bright light until I either sneeze or stop feeling like I need to. It could be that the CFL I normally use is not bright enough for my purposes. "Hazards of Light" by Cronly-Dillon et al. suggests that I would be safe using a much more intense lamp.

3atorm14y

When you say "making full use of my body", are you merely learning, as you say, "tricks", or are you also developing your muscles and body-senses (proprioception, balance) so that you can move yourself and other things as efficiently as possible?

7spqr0a114y

Both, definitely. I do parkour regularly; I can reliably run (or walk backwards) on a 7cm wide beam, and am practicing rolls. I lift weights once per week and will move to twice weekly if I want substantially more arm strength than I have now. I also hike often. I have made large gains in physical fitness; as a kid I was always near the slowest of my peers, now I am one of the fastest runners in my laser tag club.

5Crux14y

Parkour! I do it too, and Daniel Ilabaca is definitely one of my heroes. If I could say just one thing about parkour, I would say that it's not about taking thoughtless risks (linked for what he says after the jump), and not even necessarily about taking calculated ones. I personally (try to) operate under a strict maxim to never attempt any movement where I'm not trained to compensate for every possible way of messing up. I never climb anything that I can't fall off safely. Etc. This entails not just focusing on mastering the cool techniques, but also mastering all the different falling techniques to facilitate safe training. This is fine by me anyway; training the falling techniques is no worse exercise than the power moves in Ilabaca's videos. I personally am in it for the exercise and athletic development, so injuring myself would be especially counter-productive. Parkour is unlike something such as tricking, in that for every technique there's always a safe and reliable progression; it's neither dangerous nor injurious unless you practice irresponsibly and make it so. I hope to at some point develop an organized system of pre-requisites for each movement such that anybody could move through the progressions safely and reliably. That's one thing I'm working on, I suppose. The main contribution would be that it would systematize the falling techniques and perhaps invent new ones. Good parkour practitioners are generally extremely good at falling, but only because they fall a lot. It's perhaps mostly unconscious and instinctual, whereas I would make it systematic. Maybe this has been done (in which case I would appreciate a link), but I haven't seen it, and anyway parkour is a very young discipline (with a lot of risk-takers), so it would by no means be surprising if such a systematic approach doesn't exist. Anyway, I'm writing this somewhat for myself, but also for spqr0a1 and a general audience. I needed to remind myself why I do parkour and how I should ap

1atorm14y

I also do parkour, and would be interested in seeing a systematic training regimen for the relevant skills.

2Daniel_Burfoot14y

I think it would be awesome to simultaneously refine my sense of smell and my knowledge of biochemistry to the point where I could identify the actual molecules in the air. So I could detect a scent and think to myself "hey, that's a flavonoid."

2JoshuaZ14y

Extremely similar chemical compounds can have very different smells to humans and very different compounds can have very similar smells. So while this may be possible, I'm not sure that human smell senses are good enough to do this in general.

2atucker14y

This sounds awesome, but I'm fairly mystified as to why you picked that particular goal.

4Crux14y

Maybe because of how awesome it is.

2spqr0a114y

A friend of mine naturally exhibits exclusively photosensitive sneezing. So I thought it would be interesting to try. This study suggests it is primarily acquired and not inherited so I figured it was worth a shot.

2JenniferRM14y

I'd heard a while ago that the photic sneeze reflex was a so-called Mendelian trait which might actually be a "simple Mendelian trait". Having found so many other cases where such claims didn't pan out, I guess at this point I shouldn't be very surprised when another "simple Mendelian inheritance" story turns out to be complicated :-P I only read the abstract of your linked study rather than methods and everything. Out of curiosity (if you read the whole thing), did they disentangle the causal factors behind variation in the trait? Like, did they distinguish between hypotheses such as "actual non-ACHOO-carriers can acquire the reflex based on life experience" vs "only carriers can acquire the trait, but penetrance increases with age, giving the potentially false appearance of people acquiring it based purely on environmental factors"?

0spqr0a114y

I am trying to access the full article through my library system but it will take some time. It is worth noting that my goal is for light to make me sneeze, if and only if I already feel like sneezing. This is different from ACHOO syndrome as generally described; so I am unsure whether my technique uses the same biological mechanism. EDIT: Until reading your post I had not considered the possibility that I may be a carrier who had not yet expressed this trait. I thought that I would be able to acquire it through conditioning regardless. Lack of a response from my family suggests that this is an acquired trait for me. UPDATE: Indeed, http://www.ncbi.nlm.nih.gov/pubmed/7673597 was about to discriminate between this hypotheses. I have the article and it states that 39.3% of those affected reported no family history of photic sneezing. From this and other data they conclude that "... in approximately 25% of cases the [photic sneeze response] may be inherited in an autosomal dominant manner, but the majority of cases appear to be related to environmental influences." They did not identify any causal environmental factors for the formation of this response, but that it is primarily acquired is good news for anyone interested in trying it.

1Davorak14y

Do you just want to learn to control your sneezes? Or are you interested in the photosensitive effect directly? If the former I would encourage you to learn more direct control mechanism rather then using a external trigger like light. edit: spelling

2spqr0a114y

Primarily I was looking for an exercise in conditioning, any practical benefits are ancillary. If progress continues, I will not sneeze unless a specific trigger is present (staring at a very bright light); so it should be a passive benefit with no long-term upkeep. If you have better ways of control sneezing, I am interested in knowing them.

0Davorak14y

I consider it a low probability that I have enough experience/knowledge to generalize my understanding/perceptions to a wide audience with fidelity. If you want to talk about it over the phone or on skype some time I would be happy to oblige. Quick iterative discussion can do much to shorten inferential distance and if a common understanding is found easily it might be worth writing up and posting.

[-]AlexSchell14y140

I am working on finishing up a philosophy paper about whether "fine-tuning" (the claim that the physical constants and initial conditions that permit the evolution of life and conscious observers are rare in the space of physically possible parameters) supports "multiverse" hypotheses according to which the cosmos is huge and is heterogeneous in its local conditions. One major argument for the view that fine-tuning does not support multiverse hypotheses is due to Ian Hacking, who claimed that this inference is analogous to an "inverse gambler's fallacy" where a gambler enters a casino, witnesses a roll of dice resulting in double-sixes, and concludes that the people must have been throwing dice for a while.

While going through Nick Bostrom's book Anthropic Bias, I've found his discussion of Hacking's argument (and of an significantly improved recent version by Roger White, available here ) somewhat unilluminating, although I thought there must be something wrong with the argument. Going through the existing replies to this argument in the literature I've found counterarguments that either fail straightforwardly or (more commonly) render fine-tuning irr... (read more)

3antigonus14y

Have you read Darren Bradley's Multiple Universes and Observation Selection Effects? If so, do you group it into the category of unacceptably a priori arguments? Because it sounds somewhat similar, and I remember finding it convincing at the time I read it.

2AlexSchell14y

Yes, I've read Bradley's paper, and his approach is the best I've seen so far. It raises all the right questions and has been very helpful to me personally in giving me an idea of what form a plausible reply to the inverse gambler's fallacy argument would take. I do indeed think his approach collapses into an argument that is almost a priori / barely sensitive to fine-tuning (unless one adopts a fairly ad hoc metaphysical view of the necessary and sufficient conditions of your existence, a view that Bradley makes explicit in a forthcoming paper). Bradley's argument can be fixed by rejecting the methodological principle he implicitly relies on (which is the idea that the correct "selection procedure" is "biased", in his technical sense, toward your existence; Roger White also relies on this idea, which he calls the "observation principle") and replacing it with the self-sampling assumption with an at least moderately inclusive and universe-neutral reference class.

3JoshuaZ14y

I'd be very interested in reading this when you have a readable draft.

2AlexSchell14y

Thank you. I intend to post it in the discussion section with a request for feedback when I have finished the draft.

1false_vacuum14y

Yeah, this is an important subject. I'll probably read your paper. To what, exactly?

1AlexSchell14y

I fixed the second paragraph. I meant a "solution" to the challenge posed by the Hacking/White "inverse gambler's fallacy" argument, basically just an account of where exactly the argument goes wrong.

0jsalvatier14y

Also would be interested in seeing this.

[-]thomblake14y140

Been coding a seed AI. Halfway there - it's already superintelligent and has taken over the universe, just need to work on Friendliness now.

7pedanterrific14y

Ooh, I know - just ask it how to solve Friendliness. It's superintelligent, that means it knows the answer, right?

5thomblake14y

Sure, but it has no interest in telling me!

7pedanterrific14y

Have you tried phrasing the question in terms of paperclips?

[-]thomblake14y100

That's the only way I've tried phrasing the question. Why, what do we care about apart from paperclips?

2pedanterrific14y

I have no interest in telling you.

[-][anonymous]14y120

I'm working on becoming more physically attractive. Towards this end, I've been 1) upgrading my fashion based on recommendations from one of my highly skilled roommates, 2) implementing Tim Ferris' 4 Hour Body system to lose weight and gain muscle, and 3) handling random small things having to do with my physical appearance. Towards this end I've 1) bought new shirts and a watch, 2) have been eating paleo and working out twice a week (as per the recommendations from Ferris) for the last 3.5 weeks, and 3) bought new shaving cream, new facial moisturizer, hydrocortisone for eczema, cryotherapy for a wart on my toe.

My goals for the respective sub-projects are 1) new jeans, shoes, and shirts so I can wear my new style 5-7 days a week, 2) drop fat while gaining muscle until I'm under 190 for weight and have a visible 6 pack, and 3) handle the wart and have the items available to get into a constant schedule of maintaining my peak appearance in terms of skin quality and minor things like nose hair. Perhaps I'll make a weekly checklist.

I'm working on this because I made major updates this summer about the importance of physical beauty for men, for relationships and for career reasons. Se... (read more)

4Nisan14y

Are some kinds of shaving cream better than others? Does your new shaving cream have better effects than your old shaving cream?

8[anonymous]14y

The advice I've gotten is that gel is better than cream. Past that, brand hasn't been a dominant consideration for me. More important than the shaving gel was changing my shaving style - I was putting on far too much cream before. You're supposed to use a thin layer, or else the blade can skip and cut you. Since changing this behavior, I haven't been getting cut and I haven't had razor bumps. I recommend this guide, which is the source of my opinions. What I REALLY recommend, however, is this stuff. A good part of shaving correctly is making sure your pores get closed up and not getting too dried out - both will cause irritation and pave the way for acne outbreaks. With that lotion, my face feels like God, but better, because it exists.

3Nisan14y

Thanks. I think I have been shaving my neck in the wrong direction.

1jsalvatier14y

Awesome, I will have to check this out.

1Jayson_Virissimo14y

Some believe that shaving cream is a racket. Personally, I don't use the stuff.

0Nisan14y

Do you just use a dry razor blade?

7Jayson_Virissimo14y

Mach 3 + Warm Water

[-]gwern14y90

I've been working on an article/essay summarizing my work with PredictionBook.com and what I think I've learned. It's going pretty well; I think I could post it tonight, but I'd rather finish making a particular big batch of predictions first.

EDIT: It's up at http://lesswrong.com/lw/7z9/1001_predictionbook_nights/

[-]PuyaSharif14y90

Since ten years back a sub-field of quantum information theory has emerged, quantum game theory. Regard it as the intersection of quantum mechanics and game theory. It deals with game theoretical situations where the participants use entangled quantum states, quantum superposition and unitary operations as resources to gain advantages compared to classical counterparts.

I am designing and (trying) solving a quantum game using three level quantum states, qutrits (compared to the usual qubits, two level systems).

[-]SilasBarta14y60

Finding a job that will allow me not to live in Waco, my first priority. Plus, other neat stuff like the recent contest that interferes with that.

0Dr_Manhattan14y

Considering the 2 things it's hopefully a constructive interference :)

0SilasBarta14y

Hopefully indeed, though the connection is tenuous.

[-]jsalvatier14y60

I just finished learning how to type properly, which I had neglected to do in high school. I finished reading Foundations of Neuroeconomics with another LWer. We're deciding what to read next.

I'm currently trying to become competent at mathematical proofs. I've hired a tutor and I'm practicing about an hour a day. My goal is to understand probability theory as I think it will help me correct some subtle issues with a statistical package I've designed (which I am otherwise very pleased with). I'd also like to better understand some advanced MCMC techniques.

2Dr_Manhattan14y

Interested in more information about tutoring process.

0jsalvatier14y

I searched the UW math department tutor list for a tutor and selected one (there happened to be a LWer who I had met before on the list). He gives me problems to do and sections to read out of a book and looks over the problems I do, trying to correct misconceptions I have. I ask him lots of questions about doing proofs. Does that answer your question?

0Dr_Manhattan14y

Yeah. What's the rate like?

2jsalvatier14y

$50/hr there was also someone offering free tutoring in particular math topics here on LW.

0Dr_Manhattan14y

Interesting. I heard you can get skype tutoring at much smaller rates (typically from India), though not sure what the quality is.

0jsalvatier14y

That makes sense, any hints on where?

0Dr_Manhattan14y

Don't have any info beyond this, searching for "Skype tutoring rates" yield this http://www.online-math-tutor.net/skype-math-tutors-costs.htm, for example

[-]Armok_GoB14y50

I'll be doing the free online Stanford AI course.

Oh, and yesterday I were feeling to ill to do anything actually productive like hang on internet forums, so I started messing around in python and got a fair bit into an experimental, and I quote, "brain machine interface using only hardware available on a normal PC" through a bunch of neurological tricks mostly learned from this site. It's crazy enough that it might actually work, in which case it's a pretty rad shortcut to IA and getting yourself a codeic cortex for the price of only a beeping-induced aching head and a seemingly-pointless-circular-movement-induced aching hand. I may or may not work on this again.

0Swimmer963 (Miranda Dixon-Luinenburg) 14y

Cool, me too! Unfortunately I never took a linear algebra class (calculus and computer programming filled up the electives allowed to me in my program) so I will be having a crash course on it with the Ottawa LW groups tonight, and then attempting to learn the rest as I go.

[-]wmorgan14y50

I've made progress in two areas:

An intermediate goal of accumulating money & power
A primary goal of dating successfully

I situated myself as the lead for my company's sexy new product. On Tuesday I led a well-received demonstration to a potential customer, and on Wednesday I told my boss that I wanted a new title with more pay. He's a much better negotiator than I am, but we're still talking, anyway. The details of the agreement we reach or don't reach will determine where I live and work in 2012.

I continue to accumulate cash despite my best effort... (read more)

2Crux14y

Do you mind expanding on what you mean by "social games"? Maybe give some examples etc.

[-]wmorgan14y180

Sure. I'll give examples of situations where I think I've improved, rather than where I'm just failing miserably, since the failures probably mean I don't understand what's going on.

You're flirting with someone and they ask you a simple personal question, like your age or whether you're single. You can answer honestly, right away, which is what I did for years because I didn't even consider other options. Or you can make them guess. Before I started doing this, I thought they'd get frustrated and end the conversation. The opposite is true! Answering right away is boring. Making them guess makes them more invested and more interested.
More on being coy with information: you know you're going to do the dishes, no matter what. But you say to your roommate, "hey, I'll do the dishes if you take the trash out." You'll get this deal every time. But if they know upfront that you're already committed to doing the work, they'll be much less likely to do their fair share.
You're riding in a car talking with two other people. Someone slips up and says something offensive, or dumb, or otherwise immediately regretful. I think that the obvious Golden Rule response, and my go-to resp

... (read more)

1A1987dM13y

As a bonus, if you have people guess your age, you get to know how old (people are willing to admit) you look.

1Crux14y

Thanks!

[-]thomblake14y50

link to previous threads is broken. It's not set up as a relative link.

5Vladimir_Nesov14y

Fixed.

[-]djcb14y40

So, mostly to sharpen my programming mind, I am writing an e-mail client in a combination of C, emacs-lisp and guile-scheme, all based on a database (searching) approach to e-mail, but still supporting folders. I hope it'll be ready for the next version of mu. It's been especially a nice learning experience to figure out how to write non-trivial emacs-lisp programs. And I like very much how I can already use the e-mail client for my normal communications, and making improvements whenever I need something.

Then, I'm training for a marathon, still this year.... (read more)

[-]JoshuaZ14y40

I recently became aware of how much I deeply don't understand the quantum end of theoretical compsci, so I'm trying to get a decent handle on that.

I've also been doing work on some problems related to the behavior of polynomials over finite fields. Theorem: Let F_q be a finite field with q elements, and let f be a polynomial of degree n with coefficients lying in F. Then if the image of f is not all of F, then the number of elements in the image of f is bounded by q - (q-1)/n. See here for a proof.

One corollary of this is the following: Let f be a non-triv... (read more)

[-]Swimmer963 (Miranda Dixon-Luinenburg) 14y30

I've written about 50 pages in the last few weeks (working on a novel), and gotten to feel the awesome high of a writing "flow state" again, which I hadn't felt in quite a while. Schoolwork may have been neglected somewhat. I am trying to commit to at least looking at all of my stories every day and seeing if there is anything that I want to write.

I'm training for my first taekwondo tournament, which is in 2 weeks. I've made huge gains in flexibility, probably put a bit more muscle on my legs (calves especially, swimming does not do a lot for th... (read more)

[-]magfrump14y30

Improving social skills: I have made a number of new friends over the summer, took notes, started using alcohol as a social tool, and hit on and picked up a number of young women successfully.

Mathematics: I have been reading Silverman's "Arithmetic of Elliptic Curves" and attempting all of the exercises at the behest of my advisor, and attending number theory seminars at Stanford.

Practical: My computer's keyboard is having problems so I tried to fix it but I may have to replace it (every time I type "p" it also goes back a space, and ot... (read more)

0JoshuaZ14y

Most of this sounds good. The last seems less so. Lomborg makes occasionally valid points but the book is in many ways an exercise in motivated cognition.

7magfrump14y

Have you read the book? The entire premise was that he was debunking claims and ended up being persuaded by them; this seems like the opposite of motivated cognition. I won't deny that it seems fairly one sided but it does seem like taking opposing views would be difficult, since he uses the same citations as groups he disagrees with and presumably is better with statistics as a statistics professor; I don't mean to say that one should be totally convinced by his arguments, but more convinced than I should be by all the arguments without statistics I've heard before which were about 50 times as much motivated cognition.

3JoshuaZ14y

Yes. I have read it. I'm familiar with what he claims he is doing, and frankly, I couldn't tell if it was dishonesty or general lack of self-awareness about his own cognitive processes. This is not to say that he doesn't make good points on occasion. And a lot of the responses to him demonstrated about as bad motivated cognition or worse. There's a January 2002 Scientific American that had multiple essays discussing the book, some of which are worth reading, and illustrate how some of the scientists he is citing have said explicitly that he's misinterpreting their work. That's not to say that the book is a complete waste of time. There are two general points that Lomborg does make that I found to be worth-while: One, that everything has trade-offs and that much of the environmental movement doesn't appreciate that and how one should look at the marginal rate of return. I suspect that this sort of argument won't be at all original to most readers of LW. Two, that by many metrics much of the developed world is environmentally better off now than it was fifty or a hundred years ago, and by some metrics for some large cities, even larger spans of time. There are also interesting statistics he brings up which by themselves are useful for understanding a larger picture of the world. The problem with a book like this that touches on a lot of different areas where the author is both not an expert and where experts in some of the areas have specifically pointed out issues with his interpretation of their fields is that it is very hard for a non-expert to tell when he's made a good point and when he's using the data well.

2magfrump14y

Voted up. I intended to post this earlier but it didn't seem to go through: I'd be interested in references to the responses you have in mind, so that when I finish the book I could read them and incorporate them into the summary.

0JoshuaZ14y

Unfortunately, the last time I paid attention to this book was shortly after it came out, so I don't have a lot of references off-hand. The already mentioned Scientific American seems to be a good place to start (although note that when Lomborg wanted a right of response SA behaved not so well.)

0magfrump14y

Mk, well chances are it will be a while before I finish the book since I have a lot else to do at the moment; if you stumble across anything let me know otherwise I'll search myself when I get around to it.

0Crux14y

Could you explain how exactly you use alcohol as a social tool? I'm not issuing a challenge or anything; I'm just wondering what your experience has been.

6magfrump14y

Well, I looked over a handful of LW posts and PUA sites and videos and observed the key steps in socializing; something like "talk and then the other person talks and then get them to laugh and find something they're interested in and touch people in non threatening ways like on the shoulder when you talk to them." Then I looked at which of these things I had problems with, and what tools I could use to solve them; i.e. if I wore a fabulous sparkly pink hat a lot, that made it easy to get laughs at first, and if I had a few drinks first it made it easier to say the first few words to the other person. I don't know how well I'm explaining this but basically, it can be helpful to emulate being another person; on some level this has to be a brain chemistry event, and alcohol modifies your brain chemistry in an observable way. For me, that way was often helpful to my goals of being more social.

0Crux14y

I've always chalked up the useful effects of alcohol to be because it alleviates the embarrassment of making mistakes: "Wasn't my fault. I was drunk." Being too afraid of screwing up in front of others is a very common problem to have, so alcohol is helpful to a very wide variety of people. Do you think that particular effect (helpful for emulating another person) could be subsumed under my above analysis, or do you think there's something else to it?

0magfrump14y

I think that if we tried to break down which specific chemical reactions we were talking about, we would be talking about the same ones. I haven't been very eloquent about this and my keyboard is glitching at the moment so it is frustrating to try to type about it. Maybe later I could talk more about the subject but I doubt I have anything new to say.

[-]Dr_Manhattan14y20

Mostly just learning stuff. Signed up for Stanford's AI and ML courses, and just finishing Manning's "statistical natural language processing". Plan to apply some statistical methods to a rule-based system at work that needs too much maintenance.

[-]David_Gerard14y20

http://reddragdiva.co.uk/lj/nyan.mp3 - the 4yo's favourite dinosaur is the parasaurolophus, so the loved one will be coming up with an 8-bit one of those that flies through the air farting rainbows. Any good meme must have 4yo appeal.
Trying to make actually sensible sense of various concepts I've met in the worlds of aesthetics, art criticism and p-stm-d-rn-sm. After throwing various standard texts in the field against the wall with great force (well, imagining myself doing so, since I was reading them on my phone), I've been discovering that putting the

... (read more)

[-]Zetetic14y10

I've been trying to apply John Perry's Structured procrastination more purposefully. I've been taking on numerous projects that feel important but probably have negligible consequences if they are left undone.

As far as projects I've taken on (amidst having to send out grad school applications) I've been authoring a weekly op/ed column in my school paper with a friend. Our school is small enough for us to get away with just about anything with some intellectual merit; copyright issues, nootropics, the Ig nobel prizes, anything of interest that comes up. ... (read more)

[-]Jayson_Virissimo14y10

In this comment I set myself four goals. Here is a short update:

I'm teaching myself Java. My goal is to complete an app for analyzing raw genotypic data from personal genomics services (like 23andme) by summer 2012.

I'm currently 17% done working my way through Introduction to JAVA Programming, by Y. Daniel Liang.

I'm training 6 days a week to score 300 points on the USMC PFT (I have no intention of joining the military, I simply think it is a good goal in terms of overall physical fitness). I am increasing my running speed at a rate that is 75% slower

... (read more)

3[anonymous]14y

What do you mean by folk activism?

1Jayson_Virissimo14y

This.

0[anonymous]14y

GREAT article, thanks for linking. This isn't stressed enough.

[-]BenLowell14y10

Last time I said:

So, I've started writing an article related to this in collaboration with another LWer. One of my goals is that like the idea of contributing content to the site. I was also curious, and I feel that the majority of my personal curiosity is satisfied, and finishing the project by communicating what I learned to others is what is being difficult. There are several reasons for this. One is that I can always learn more, and make a better article. I've also realized this is a lower priority than my school/work activities, so it keeps being pu

... (read more)

[-]anotheruser14y10

I have been trying to invent an AI for over a year, although I haven't made a lot of progress lately. My current approach is a bit similar to how our brain works according to "Society Of Mind". That is, when it's finished the system is supposed to consist of a collection of independent, autonomous units that can interact and create new units. The tricky part is of course the prioritization between the units. How can you evaluate how promising an approach is? I recently found out that something like this has already been tried, but that has happen... (read more)

[-]jsalvatier14y130

I suspect some people here will have a negative reaction to your comment. Your approach comes off as not very serious, your last paragraph sounds like reasoning from conclusion to argument, and your mention of friendliness seems like an afterthought.

1anotheruser14y

I assure you that I have thought a lot about freindliness in AI. I just don't think that it is reasonable or indeed possible to make the AI have a moral system from the very start. You can't define morality well if the AI doesn't have a good understanding of the world already. Of course it shouldn't be taught too late under any circumstances but I actually think that the risk will be higher if you try to hardcode friendliness into the AI at the very beginning, which will necessarily be flawed because you have so little to use in your definition, and then work under the assumption that the AI is friendly already and will stay so, than if you only implement friendliness later once it actually understands the concepts involved. The difference would be like between the moral understandings of a child and an adult philosopher.

7Vladimir_Nesov14y

Have you read a good AI/machine learning textbook, like AIMA or shorter Mitchell's book? Let your goal drive you to study and learn and refine yourself and become stronger.

3anotheruser14y

I read the first one, but it didn't really cover learning in a general sense. The second one sounds more interesting, I wonder why I haven't heard of it before. Do you know where I can get it? I'm a student and thus have very little money. I don't want to spend 155$ only to find out it only contains stuff I already read elsewhere.

5Vladimir_Nesov14y

OK, if you've read AIMA and still want to become a Dark Lord, I don't know if I should encourage you on this path. My impression is that Mitchell's textbook covers less material than AIMA, though I didn't read AIMA.

1anotheruser14y

What gives you the impression that I "want to be a Dark Lord"? I have already explained that I realize the importance of friendliness in AI. I just don't think it is reasonable to teach the AI the intricacies of ethics beore it is smart enough to grasp the concept in its entirety. You don't read Kant to infants either. I think that implementing friendliness too soon would actually increase the chances of misunderstanding, just like children that are taught hard concepts too early often have a hard time updating their believes once they are actually smart enough. You would just need to give the AI a preliminary non-interference task until you find a solution to the friendliness problem. You might also need to add some contingency tasks such as "if you find you are not the original AI you but an illegally made copy, try to report this, then shut down.".

3Vladimir_Nesov14y

1jsalvatier14y

Searching booksprice.com yields a used version for 40 usd. You can also find a lot of books online through torrents and such.

0anotheruser14y

thanks for the effort but I just found out that the library at my university does have the book after all. I overlooked it at first because the search engine of the library is broken.

1pedanterrific14y

Somehow "DANGER WILL ROBINSON" doesn't seem to quite cover it.

[-]James_Blair14y00

A programming language that has the semantics necessary to elegantly express a new kind of garbage collector. A rudimentary prototype of the collector, written in C++ with a terrible interface, appeared to confirm the idea.

At least, that was the initial goal. The more I investigate the design choices involved in programming languages, the more room for improvement I see.

1Shmi14y

Just wondering what "new kind" you mean, assuming you are OK discussing it. The last language with a decent garbage collector I came across was Lua, and it has a fairly granular control for garbage collection in v5.1.

2James_Blair14y

Like a reference counter, it doesn't need to run collection cycles as it knows exactly when memory can be freed. Despite this, it handles circular structures just fine. Apart from that, there's nothing special about it. The idea was absurdly low hanging fruit and that aspect might make it good LW post material. Assuming that it works as well as I think it does.

Moderation Log