Louie Helm: What I Learned From Less Wrong
For those curious about what all the fuss over Less Wrong is about, Louie Helm, formerly a Singularity Institute Visiting Fellow, has a post up on what he's learned from reading the core sequences and following it up with participation.
500 Pictures of Singularity Summit 2010 Available
Photo by A. Jolly 2010.
My Flickr account contains almost 500 photos of Singularity Summit 2010, more than you could ever want. I mentioned this before in the Singularity Institute newsletter but not here. A special thanks to our photographers, A. Jolly and Anthony Scatchell. Please get in touch with me if you are interested in volunteering for photography next year.
Steven Mann is so cool!
The videos are currently being edited, they'll be completed over the next few weeks. Sorry for the delay, one of our initial editors backed out of the project. Watch the Vimeo channel for updates. I'll announce it officially on SIAI blog when some go online.
Total attendance at Singularity Summit 2010 was approximately 620.
For any new readers: the Singularity Summit is put on by the Singularity Institute, which I work for. I co-organize Singularity Summit, assisting our President, Michael Vassar. Everyone at the Singularity Institute cooperates to make the Singularity Summit happen. The Singularity Summit MC is Sean McCabe, previously a close assistant to James Randi.
New America Foundation’s “Never Say Die” Conference
You can fast forward to 31:00 for Aubrey. The conference website is here.
Moderator: Explain what SENS stands for.
Aubrey: No, it's too long...
Resveratrol doesn't extend lifespan. The moderator must not have gotten the memo as she brings it up as a potential life-extending compound.
Obama’s Bioethics Panel and the Transitory Nature of Presidential Bioethics Panels
I've been reading some of the public material on the new President's Council on Bioethics, because I can't for the life of me figure out what they do. I do know that they're called the "Presidential Council for the Study of Bioethical Issues" now. Same domain name, different panel. It seems as if Obama was so against the old commission that he had to destroy it and create a new one from scratch, which highlights the transitory and low-power nature of the body.
Checking out the background materials section of their website, I was compelled to click on the first presentation at meeting two, "Oversight of Emerging Technologies". It outlines important overall characteristics of this panel. Their mission is as follows:
1. To monitor scientific/medical developments (“advancesâ€) and identify the issues they will raise for society
2. To bridge divide between science and society
3. To articulate the range of views on controversial subjects, To inform the political process & policymaking
4. To provide guidance to individuals & healthcare professionals
5. To provide recommendations to policymakers
Another important part, under "mode of work", is this:
Not asked to invent new philosophical theories but to offer conclusions & recommendations based on multidisciplinary analysis of issues facing policy makers, healthcare professionals, scientists, patients & families
I love this part. It seems to be a nod to the theological conservative philosophy underlying Bush's panel, and backing away from that attitude. Some of the panel members are no doubt aware of transhumanism as well, so the statement might be seen as a reassurance that the panel won't take sides and align with any particular philosophy.
The problem with not inventing or using existing philosophical theories is that you disempower yourself. Philosophical theories often dictate the course of history. The Presidential panel's decision not to embrace philosophies for guiding bioethical decisions (as if that's even possible) creates a power vacuum for transhumanists and bioconservatives to fight over.
Because the panels are so transitory and ephemeral anyway, they lack stability and power. Transhumanism, in contrast, is an ongoing, capable, increasingly higher-profile community that was more or less founded with the launch of the extropians mailing list in 1991. Bioconservatives, meanwhile, mostly focus on near-term issues like abortion, outlawing marijuana, and assorted anti-gay bigotry. With regard to future issues, the issues that will determine the trajectory of the 21st century, they are mostly highly disorganized or silent. The New Atlantis, the bioconservative journal, has received barely any external coverage in its seven years of existence. Surprisingly, sometimes the editor's column in the Wall Street Journal more closely resembles transhumanist articles than anything written by Leon Kass.
As far as I can tell, the main function of Obama's panel seems to be to assess the potential risks of synthetic biology, which is fantastic. The phrase "life extension" does not appear anywhere on the site.
Superior Retinal Prosthesis Developed for Mice
From Science Daily: A new retinal prosthetic creates an image (middle) that more accurately reconstructs a baby's face (left) than the standard approach (right).
Researchers have developed an artificial retina that has the capacity to reproduce normal vision in mice. While other prosthetic strategies mainly increase the number of electrodes in an eye to capture more information, this study concentrated on incorporating the eye's neural "code" that converts pictures into signals the brain can understand.
Degenerative diseases of the retina -- nerve cells in the eye that send visual information to the brain -- have caused more than 25 million people worldwide to become partially or totally blind. Although medicine may slow degeneration, there is no known cure. Existing retinal prosthetic devices restore partial vision; however, the sight is limited. Efforts to improve the devices have so far largely focused on increasing the number of cells that are re- activated in the damaged retina.
This is a major BCI advance. Prior visual reconstruction implants had a much lower resolution. Within a couple decades it could become possible to use implants like this to generate and share complex virtual realities just "beamed" into one another's heads. Ramez Naam's More Than Human describes a similar technology and examines how it could be used to enhance human collaboration and the creative process.
If you can reconstruct a real scene and beam it to the brain, then you can also produce fake scenes if you have the right programming.
Zyvex Labs: “Atomic Precision Fabrication Using Patterned Si Atomic Layer Epitaxy”
Here's a description of what they're doing. Zyvex is the best-funded group working towards molecular nanotechnology. There is good coverage of Zyvex at Next Big Future, including an interview with CEO Jim Von Ehr from May. Here's an answer I thought had interesting details:
Question 6: Eric Drexler has advocated a DNA origami approach, but others favor a direct to diamondoid strategy. Which approach do you favor?
Answer 6: We actually have our own distinct approach which is neither DNA origami nor direct to diamondoid. The ease of programming a computer controlled milling machine, which could make all manner of macro-scale products out of metal or plastic simply by changing the program, makes our paradigm compelling if we can build something similar at the nanoscale. The DNA approach doesn't lend itself to that flexibility very well. The diamondoid approach may be a great end point, but we simply don't have that capability now. We lack the precision and well defined tips to do diamondoid. By contrast, our approach gets us to rudimentary molecular manufacturing fairly quickly.
A recent presentation by John Randall (VP of Zyvex Labs) at AVS 2010 is available here. Jim Von Ehr predicts rudimentary molecular manufacturing by 2020.
Singularity Hub Posts on the Imaginary Foundation Art Show in SF Tonight
Here's the link. Myself, Michael Vassar, and various other SIAI folk will be there.
David Brin on Existential Risk and the Lifeboat Foundation
From the educational videos section of the Lifeboat Foundation website.
Starcraft AI Competition Results Posted
Starcraft AI is jacked up and good to go! Last month UC Santa Cruz's Expressive Intelligence Studio ("exploring the intersection of artificial intelligence, art, and design") held the Starcraft AI Competition, where bots were pitted against each other and human players. 29 teams submitted a bot. They played with the Brood War add-on.
This competition is interesting to me because 1) Starcraft: Brood War is my all-time favorite multi-player game, 2) it's many times more complicated than chess or Go, 3) the game requires realtime decision-making skills, 4) the best known strategies are highly complex, involving extensive micromanagement of individual units. Some professional Starcraft tournament players input hundreds of commands per minute. A medium-level player like myself probably inputs 20-30 moves per minute as the game starts to pick up. One way to win the game easily against novices is to take the optimal route to mass-producing the cheapest unit (this is known), then rushing the enemy base. This can only be done with extensive micromanagement, but once you know how to do it, it is easy. What I'm most interested in is not AI that can achieve that (the AI that comes with the game can), but AI that can do better in later gameplay, where dozens of different units with unique abilities are introduced and there is a combinatorial explosion of possibilities.
In the end, the champion human player, =DoGo=, a World Cyber Games 2001 competitor, beat all the AIs, but apparently it was close. This is impressive because an AI can simultaneously "pay attention" to all units on the map at once while a human player has to remember everything and can ultimately only focus on one unit at a time, even if it's only for a fraction of a second. Kotaku Australia has coverage of the event, but the real coverage is at the EIS blog, which has detailed technical results on all the matches.
The More We Talk, the Less We Might Agree: Study Shows Discussion Can Hurt Consensus-Building on Science/Technology
From Nanowerk News:
When it comes to public issues pertaining to science and technology, "talking it out" doesn't seem to work. A new study from North Carolina State University shows that the more people discuss the risks and benefits associated with scientific endeavors, the more entrenched they become in their viewpoint – and the less likely they are to see the merit of other viewpoints.
"This research highlights the difficulty facing state and federal policy leaders when it comes to high-profile science and technology issues, such as stem cell research or global warming," says Dr. Andrew Binder, an assistant professor of communication at NC State and lead author of the study. "Government agencies view research on these issues as vital and necessary for the country's future, but building public consensus for that research is becoming increasingly difficult."
The researchers set out to see how people talk about risks associated with unfamiliar science and technology issues, Binder explains. "Most people, when faced with an issue related to science and technology, adopt an initial position of support or opposition," Binder says. "Our results demonstrate very clearly that the more people talk about divisive science and technology issues, the less likely the two camps are to see the issue in the same way. This is problematic because it suggests that individuals are very selective in choosing their discussion partners and hearing only what they want to hear during discussions of controversial issues."
In the study ("Interpersonal Amplification of Risk? Citizen Discussions and Their Impact on Perceptions of Risks and Benefits of a Biological Research Facility"), the researchers focused on public debate related to the National Bio- and Agro-Defense Facility (NBAF), which the federal government discussed building in one of six sites around the country. Some members of the public opposed building a facility housing highly infectious animal diseases in their community. The six proposed sites were Athens, Ga., Manhattan, Kan., Plum Island, N.Y., Butner, N.C., Flora, Miss., and San Antonio, Texas. Manhattan was ultimately selected as the site for the NBAF.
The researchers conducted surveys of residents living near the proposed sites to collect data on people's perceptions of the potential risks and benefits associated with NBAF. Specifically, the results showed that, among people who opposed the facility, the more an individual discussed the issue with other people in their community, the more firmly entrenched he/she became in his/her perception of greater risks and fewer benefits. Conversely, among those who supported the facility, increased discussion led to an increased perception of benefits and a decreased perception of risks.
This research was done as part of an overarching grant project funded by the National Science Foundation, which is aimed at understanding the public opinion and policy dynamics surrounding site-selections for federal research facilities.
"This work will likely inform future decision-making on how federal agencies engage the public in regard to large-scale research initiatives," Binder says.
Source: North Carolina State University
Want to be an effective technology activist? If this study is right, maybe you should only talk to people strongly predisposed to agreeing with you from the start, or just develop the technology and forget about talking about it all day. (Some readers may be shocked to hear that this is the default with most companies, which don't actually require your explicit permission to develop and commercialize novel technologies. Amazing, I know.)
New Numenta Paper: “Hierarchical Temporal Memory”
It's available here, thanks to Sean O'Neill for the link.
Long-time readers of this blog will recall that I'm unimpressed with Numenta, but I'm still interested in seeing what some other people think is a promising route towards AI. Skimming through the paper, the first 20 pages (at least) seems to be basic neuroscience.
Numenta seems to be based on a variant of the thinking behind the failure that is most neural networks. The thinking behind neural networks seemed to be, "if we superficially copy this one aspect of the brain (nodes and links, essentially), we'll get more intelligent machines!" Numenta says, "if we superficially copy hierarchical networks of neurons under a prediction-driven architecture with sparse distributed representations, we'll get more intelligent machines!" I don't see the fundamental difference.
From "neural networks" on Wikipedia:
A. K. Dewdney, a former Scientific American columnist, wrote in 1997, "Although neural nets do solve a few toy problems, their powers of computation are so limited that I am surprised anyone takes them seriously as a general problem-solving tool." (Dewdney, p. 82)
Many people in AI are so confused that popular enthusiasm can help dictate the course of the field. Neural networks are one of those fields that can easily inspire wonder among many different people, but has very limited practical value towards AI. Conversely, probability and decision theory approaches are an area of AI research that practically no one outside the field has heard of, but which seems deeply crucial to long-term AI success.
(Update: Changed "decision theory" to "probability and decision theory approaches".)