SIAI Selected as Mentor for Google Summer of Code 2009
From SIAI Director of Open Source Projects David Hart:
We are happy to announce that the SIAI has been selected again this year to participate in the Google Summer of Code program as a mentoring organization. GSoC is an annual program that awards successful student contributors a 4500 USD summer stipend to work on open source and free software projects for three months. Around one thousand students worldwide participated in GSoC 2008, with eleven students working on OpenCog related projects. Students may apply for GSoC 2009, beginning at the SIAI organization page. The student application period closes on April 3, 2009 at 19:00 UTC.
See some current OpenCog projects.
Progress Towards Haptic Suits
On this blog I've written about haptic suits before. Particularly, I think that people will use them for cybersex with bots and other people as soon as they are technologically possible. They'll make martial arts less painful by lowering the strength of impacts via cyber-fighting. Users of advanced haptic suits be able to engage in martial arts tournaments for as long as their stamina holds out, or participate in online RPGs where magic and virtual swords "exist" and can be felt. As I said in the linked post, "I predict that convincing haptic suits will arrive by 2020." Haptic suits will never be perfect, though. You can't genuinely simulate getting hit with 100 lbs of weight if the suit itself doesn't weigh 100 lbs.
Today I saw a press release on a primitive haptic jacket, that "can enable movie viewers to feel movies through a sense of touch, in an attempt to provide full emotional immersion in a film". This was presented at the IEEE-sponsored World Haptics Conference 2009, in Salt Lake City. Here's the program if you're interested in browsing.
The Brain’s Virtual Reality
The human brain has a fuzzy virtual reality module, which kicks into action during dreams and serious drug trips like LSD and DMT. This VR module can render color 3D scenes, often with physically unrealistic characteristics (unless you're alexithymic, as we discovered in Ian's comment in the last post), with objects that can morph and rapidly appear and disappear. It does seem to be difficult to focus on small details in these alternate scenes, however, especially the reading of text, whether real or in a dream.
My guess is that these "VR scenes" are just perturbations of the usual virtual scene which our brain creates that we call "reality". Based on evolutionary conservation of complexity, this seems like the most likely possibility.
In the future, it might become possible to hijack the way this "personal VR" works with brain-computer interfacing. That's what I mentioned in the post about brain-computer interfaces for manipulating dreams and what Kurzweil calls "experience beaming" in his books.
From our current vantage point, we call dreams and hallucinogen trips "weird" because they reflect external reality less accurately than sober awareness, and introduce unlawful and odd complexity that doesn't help the observer manipulate or understand the true external environment as well. (It does give them another unique cognitive reference point which may have its own special benefits, but nothing too useful. Some people claim to have lucid dreams, and people have come to legitimate epiphanies about their lives or the world while on certain drugs.)
From the perspective of a posthuman with a different "experiential palette", or own daily experience may seem like a silly drug trip, a drunken blur. For instance, distant objects look small and undetailed to us. A posthuman with direct experiential connections to sensors throughout the environment might be able to behold the full features of the distant object via a light-speed connection to remote sensors. Isn't it more natural to behold as many things as possible in their full size and detail, rather than be restricted to objects immediately around oneself and in one's line of sight?
If there are such major perceptual shortcomings that we have, then isn't it possible that they might be accompanied by cognitive or imaginative shortcomings that we lack even the self-reflection to perceive?
Are Emotions Necessary for General Intelligence?
From where I'm standing philosophically, the answer is "obviously not, our particular emotions are contingent aspects of human intelligence which exist for specific evolutionary reasons". In trying to find more evidence for this opinion, I found this Wikipedia article on alexithymia:
Alexithymia (pronounced /əˌlÉ›ksəˈθaɪmiÉ™/) from the Greek words λÎξις and θÏμος modified by an alpha-privative — literally "without words for emotions" — is a term coined by psychotherapist Peter Sifneos in 1973 to describe a state of deficiency in understanding, processing, or describing emotions.
The formal definition is here:
1. difficulty identifying feelings and distinguishing between feelings and the bodily sensations of emotional arousal
2. difficulty describing feelings to other people
3. constricted imaginal processes, as evidenced by a paucity of fantasies
4. a stimulus-bound, externally oriented cognitive style.
Continuing down the page, it looks like some pop-psychology BS is being deployed to define this category, but could it still be valid? If someone really has "no emotions", it seems like they'd need to have some genetic disorder that somehow suspended functions in parts of the limbic system without killing them entirely. It seems possible to me that alexithymiacs just have lower emotional arousal than others, rather than the genuine absence of all type of emotion -- but who knows?
Some initially pop-psych sounding categories do end up seeming legitimate. Like psychopaths, for instance, which truly do have different neural activation patterns than normal people when exposed to things like violence.
Is anyone else familiar with alexithymia or any empirical research on it?
NYT: The Conficker Worm: April Fool’s Joke or Unthinkable Disaster?
Interesting article at the NYT about the Conficker worm:
Given the sophisticated nature of the worm, the question remains: What is the purpose of Conficker, which could possibly become the world’s most powerful parallel computer on April 1? That is when the worm will generate 50,000 domain names and systematically try to communicate with each one. The authors then only need to register one of the domain names in order to take control of the millions of zombie computers that have been created.
One fictional example of a massive cyber-disaster I've heard of is *****'s Kiss, but I haven't actually seen the show it's on, .hack.
Massive Botnet: coming to a computer near you April 1, 2009.
PGD-IVF Would Lead to Designer Babies
George Dvorsky at Sentient Developments points us to an op-ed at New Scientist titled "Fears over 'designer' babies leave children suffering". The author writes:
Such fears are misplaced: IVF-PGD is little use for creating designer babies. You cannot select for traits the parents don't have, and the scope for choosing specific traits is very limited. What IVF-PGD is good for is ensuring children do not end up with disastrous genetic disorders.
I, along with dozens of prominent scientists in the field, disagree -- IVF-PGD would be useful for creating designer babies. Would would would. To boost this position, the author links another New Scientist article... (one that he probably edited, being biology features editor) which seems to contradict him:
Part of the problem is that only one or two cells are available for screening. Until recently this greatly restricted the tests that could be done. However, new ways of amplifying DNA are making it possible to do hundreds of tests. That means clinics will be able to screen for a much wider range of harmful mutations - and for desirable variants too.
Only one paragraph that I can find appears to support the op-ed author's idea that IVF-PGD couldn't be used for designer babies:
How much further can selection go? What of that object of tabloid hysteria, the "designer baby"? Will we one day be able to ask for a tall, musical, blue-eyed boy or a dark-haired girl? Even if regulatory authorities allow us to use PGD to select desirable gene variants, there are major snags. For starters, IVF typically generates fewer than 10 embryos per cycle. This means parental choice will be very limited. "I don't think anyone in their right mind would ever go through IVF to select the hair colour of their offspring," says Yuri Verlinsky of the Reproductive Genetics Institute, Chicago, one of the pioneers of PGD.
This Verlinsky quote is really confusing. Elsewhere, Verlinsky has been quoted as saying that PGD-IVF could lead to a "disease-free society" (a sloppy way of saying a "genetic disease free society"), but he claims that people won't use it to choose the hair color of their offspring. His quote doesn't make it clear whether he's talking about his opinion or the technical challenge. Also, the author of that (non op-ed) piece seriously breaks journalistic neutrality by calling designer babies an "object of tabloid hysteria" when many prominent scientists in IVF take the idea seriously. It's like the contributors at New Scientist are on misguided vigilante missions to make emerging technologies sound more palatable to the mainstream.
In any case, the limitation on the number of blastocysts can be circumvented using multi-generational in vitro embryo selection, which Verlinsky should have already considered, and if he hasn't, he has tunnel vision. So he either 1) is scientifically uncreative in his own field, or 2) knows that more advanced PGD-IVF could be used for designer babies, and just wants to keep it a secret from the public because he wants to get them to accept the technology incrementally, like boiling a crab in water that increases in temperature only slowly.
In general, I think the op-ed is a shoddy example of memetic engineering -- the author is trying to distract attention away from the designer baby controversy to help promote PGD-IVF for eliminating genetic diseases. Good motive, but somewhat dishonest, because I doubt that even the author believes that PGD-IVF would be useless for designer babies.
Speaking of "designer babies", I hate the term. As James Hughes said in WIRED, "the term "designer babies" is an insult to parents, because it basically says parents don't have their kids' best interests at heart". How about just "PGD-IVF babies", a non-catchy term, because it shouldn't become catchy and be used to discriminate against children born using the tech or parents who decide to use it? This would be in the same vein of Aubrey calling his project "Strategies for Engineered Negligible Senescence" to make it deliberately difficult to misunderstand. The project of SIAI could become, "the Engineering of a Human-Values Reflective Optimizing Process".
Also, we must remember that New Scientist lacks credibility. Instead of reading New Scientist, how about PhysOrg and Eurekalert? Or Next Big Future?
Either way, the whole issue matters not, because designer babies are largely irrelevant and will be eclipsed by things like strongly self-improving superintelligence and molecular manufacturing. See "Evolution by Choice" by Mitchell Howe.
Slashdot is Dead
Every so often, I run into someone who thinks that Slashdot still matters. Just FYI, Digg overtook Slashdot over three years ago, and for Slashdot, it's been downhill from there. Look at the current picture on Alexa:
From the perspective of sites like Digg, Slashdot barely exists. It's so bad that in January, this very website practically surpassed the traffic of Slashdot, at least according to Alexa. And you can't get much more pathetic than that.
(Update: a commenter points out what's already been pointed out many times, which is that Alexa isn't perfect, then tries to defend Slashdot on that basis, which is like a flashback to 2006. But the alternate tracking service he cites, Compete, still demonstrates that Digg is more than 40 times more popular than Slashdot, which proves my point that Slashdot is dead, and has been for a long time.)
H+ Magazine Website is Online
The people at H+ made a cool website to go along with their magazine. Check it out:
The editor, R.U. Sirius, has promised daily updates. Looking in the articles section, I see a summary of the recent AGI-09 conference by Ben Goertzel, a person whose sheer textual output probably exceeds 100 typical transhumanists combined.
There is a community section:
R.U. Sirius has a blog with one amusing post on Flash magazines. I can identify with all the points he makes.
Love the opening line. The slightly jokey/condescending tone throughout the article indicates something... what could it be? Personality.
The Terrible, Horrible, No Good, Very Bad Truth About Morality
I just finished Joshua Greene's paper, "The Terrible, Horrible, No Good, Very Bad Truth About Morality and What To Do About It" and loved it. Best paper I've read since Omohondro's "Basic AI Drives". If people just read and understood both those papers, then far more would probably drop everything they're doing and do something about unfriendly AI. I hope that SIAI can build some sort of relationship with Greene, or get him to say a few words on how moral revisionism (in the sense of moving past moral realism) applies to AI morality.
Though the first half of Greene's essay was the most intellectually serious and useful, the latter half was more amusing and interesting. I want to comment on it, but I don't want to ruin it by having you reminded by the part I mention for the whole essay, so I'm just going to post the quote he concludes with:
“If only it were so simple! If only there were evil people somewhere
insidiously committing evil deeds, and it were necessary only to separate
them from the rest of us and destroy them. But the line dividing good and
evil cuts through the heart of every human being. And who is willing to
destroy a piece of his own heart?â€
—Aleksandr Solzhenitsyn
Study gives more proof that intelligence is largely inherited
From PhysOrg:
They say a picture tells a thousand stories, but can it also tell how smart you are? Actually, say UCLA researchers, it can.
In a study published in the Journal of Neuroscience Feb. 18, UCLA neurology professor Paul Thompson and colleagues used a new type of brain-imaging scanner to show that intelligence is strongly influenced by the quality of the brain's axons, or wiring that sends signals throughout the brain. The faster the signaling, the faster the brain processes information. And since the integrity of the brain's wiring is influenced by genes, the genes we inherit play a far greater role in intelligence than was previously thought.
Genes appear to influence intelligence by determining how well nerve axons are encased in myelin — the fatty sheath of "insulation" that coats our axons and allows for fast signaling bursts in our brains. The thicker the myelin, the faster the nerve impulses.
Thompson and his colleagues scanned the brains of 23 sets of identical twins and 23 sets of fraternal twins. Since identical twins share the same genes while fraternal twins share about half their genes, the researchers were able to compare each group to show that myelin integrity was determined genetically in many parts of the brain that are key for intelligence. These include the parietal lobes, which are responsible for spatial reasoning, visual processing and logic, and the corpus callosum, which pulls together information from both sides of the body.
The press release mentions the possibility of enhancing human intelligence:
And could this someday lead to a therapy that could make us smarter, enhancing our intelligence?
"It's a long way off but within the realm of the possible," Thompson said.
I guess it's becoming more acceptable for mainstream scientists to talk about human intelligence enhancement. I wonder if the increasing profile of transhumanism has anything to do with that, or if it's about more of these ideas in fiction, or something else.
Hesperonychus elizabethae
Discovery of new North American dinosaur. Loved the artists' illustration.