Accelerating Future

SIAI stands for “Singularity Institute for Artificial Intelligence”, a educational and research group centered around the concept of the “Singularity”. The term Singularity is used to describe a distinct event - the creation of an intelligence smarter than Homo sapiens. Basically, an intelligence significantly smarter than any human genius, past or present. The Singularity Institute is attempting to trigger this event by serving as a magnet for people interested in contributing money to pay people to work full-time on the task of constructing a true Artificial Intelligence capable of improving its own source code in an open-ended way, without help from programmers.

All humans are members of the same species, with the same basic mental hardware. Our panhuman set of mental hardware is comparable to the panhuman set of physical hardware - arms, legs, muscles, organs, etc. These pieces of hardware may be larger or smaller, slightly faster or slightly slower, but share the same basic features and characteristics. Our ability to imagine and solve problems is fundamentally limited by our mental hardware. For example, it is impossible to hold more than 8 or so items in our working memory while solving a problem. We simply do not have the capacity. It is impossible for us to intuitively imagine complex, interconnected relationships beyond a maximum number of nodes. And the list goes on.

Homo sapiens is the first species capable of building a complex civilization. But it doesn’t mean we are the smartest type of mind that could potentially exist. In the possibility space, there must exist minds smarter than us as we are smarter than chimps. Problems which seem impossible to the smartest human geniuses might appear utterly simple to these smarter intelligences. A chimp genius is no match for an average human, and a human genius is no match for a genuinely smarter species.

Smartness is the quality that makes it impossible to write a story containing a character smarter than you are. You are truly incapable of imagining what they’d do. You can use crude literary devices like saying they’re capable of learning dozens of languages in a year, or memorizing 1000 digits of pi, but imagining the more subtle consequences of heightened intelligence is impossible.

Where do we find an intelligence smarter than humans? We could evolve one, by forbidding humans below a certain IQ level to have children. However this would take many thousands of years and would curtail fundamental human freedoms. We could wait for smarter-than-human aliens to arrive on this planet… but there is no evidence of aliens visiting this planet in its multi-billion year history and it doesn’t seem likely that they’ll be stopping by our neighborhood anytime soon. So the next best option is to create one with technology.

All of the above is very difficult for a lot of people to process. Imagining something smarter than us is truly difficult. We tend to think that human geniuses represent the upper ceiling of what is theoretically possible. And when we imagine smarter-than-human intelligences, we are liable to underestimate the true novelty of the prospect. Understanding our particular human type of intelligence requires exposure to some amount of cognitive science.

The Singularity Institute is attempting to build an intelligence entirely outside the human realm, through the route of Artificial Intelligence. There are a few reasons why building an AI is likely to be simpler than commonly thought. First, an AI needn’t duplicate the full complexity of the human brain. Human intelligence evolved relatively recently, and most neurological complexity exists to facilitate all the survival instincts of animals we evolved from. Secondly, human intelligence is just a particular implementation of intelligence, designed blindly by evolution rather than purposefully by an intelligent designer. A plane is not as complicated as a bird. Thirdly, the underlying hardware - silicon - is inherently more flexible, reprogrammable, and rapid than neurons.

Greater intelligence, coupled together with the right initial motivations, could help humanity more than we can help ourselves. Some would argue that any smarter intelligence would inevitably see us as inferior. The Singularity Institute argues that this is a misconception based on the way humans are programmed by evolution to interact with each other.

Compared to the posts below, this is the longest and most confusing. That’s because the topic is so difficult to discuss because there are so many facets to it. However, I believe that working towards the creation of human-friendly, superhuman intelligence represents a humanitarian cause greater than any other. And I believe that the creation of a human-unfriendly superhuman intelligence represents a risk greater than any other. For more information, see the Singularity Institute’s website.

5 Responses »

The Haber-Bosch process, mastered in the First World War, is a chemical method for the mass synthesis of fixed nitrogen - the kind plants can use as fertilizer - from the nitrogen in the air and readily available hydrogen. Before this technique was developed, massive amounts of nitrate (mostly bat dung) from Chile was shipped to farms everywhere in the world to meet fertilizer demand. As demand overwhelmed supply, scientists began to search for a way to mass produce fixed nitrogen, and the Haber-Bosch process was invented. Today, the Haber-Bosch process is used to produce more than 500 million tons (453 billion kilograms) of artificial fertilizer per year; roughly 1% of the world’s energy is used for it, and it sustains about 40% of Earth’s population.

Without the Haber-Bosch process, over a billion people, the vast majority presumably leading lives worth living, would simply not exist. Food would be too expensive. This process, however, enabled the mass production of cheap food which is foundational for population growth and health, which in turn leads to economic development. If the Haber-Bosch process were invented a couple decades later, today’s 2005 might more closely resemble the technology and prosperity level of 1995. The inverse is also true - if the process were invented before the turn of the century, we would be a lot further along by now.

The Haber-Bosch process was initially developed because the First World War made it impossible for Germany to get nitrates from Chile. At this time, the German government began to generously fund Fritz Haber’s laboratory. Haber (with the later help of Carl Bosch) solved the government’s nitrate problem.

When the telephone became widespread, it saved companies many billions of dollars per year in courier costs. The invention of the automobile allowed the expansion of society and personal freedom on an unprecedented scale. The benefits of the Internet and email are massive and difficult to quantify. Certain vaccines have saved millions.

There are technologies right around the corner, that, if their development were specifically accelerated, the benefits would greatly exceed any of the immediate costs. These are technologies that, if correctly administered, would vastly improve quality of life for everyone.

The costs and benefits curve of a revolutionary technology generally looks like the following:

The first step occurs when a genius realizes the feasibility of the technology and starts envisioning the prerequisites for its realization. The genius is then joined by many others - fellow collaborators, investors, supporters, funding apparatus. There is a snowball effect of investment, until that magical moment when the technology is deployed and the payoff begins. The near-term benefits can be enormous. Compare the initial investment in Google to its market capital today. The technology provides profound benefit, undergoes several rounds of improvement, and the level of usefulness it confers ultimately levels off - but can remain high indefinitely. (We still use the light bulb even though it hasn’t undergone any huge improvements lately.)

So which technology should we be investing in today, in order to reach the greatest possible benefit in the shortest amount of time? In rough order:

1. seed AI
2. molecular manufacturing
3. space beanstalk
4. solar towers
5. orbital solar satellites
4. space elevators
5. fusion reactors
6. germline engineering
7. handheld water purification
8. $1000 genome
9. $100 laptop
10. hydrogen fuel cells

Further information on all the above can be found on Google. The order is determined by a combination of cost, development time, and potential benefit. Seed AI totally outscores all the others, because there is no other technology on the list that itself can produce further technology, or assist in ameliorating technological risk.

6 Responses »

The above is an artists’ rendering of Burj Dubai, a tower that will measure 700 - 900m in height (the tallest on Earth) upon its completion in 2008. Its location is the city of Dubai in United Arab Emirates, home to several other artificial wonders. Construction began only recently, with around 20 floors completed so far. The exact height is being kept a secret by developers. The cost is approximately $8 billion.

Why build towers so tall? Because we have the technology! The top will sway back and forth up to about a dozen feet, which is typical for buildings of this size, but unnoticeable because it’s so gradual. Not only will Burj Dubai will be the tallest occupied building on Earth, it will be the tallest manmade structure of any kind (including radio towers). It will not even be beaten by the proposal for a Solar Tower in Australia, because recent developments have reduced its planned height from 1000m to 650m.

700m above the ground, the distance to the horizon is approximately 65 miles, compared to 3 miles to the horizon at ground level. The field of view increases from 5 square miles to 100 square miles.

Skyscrapers have appeal besides the wow factor. Making use of vertical expansion allows us to condense a lot of activity, services, and interactions within a smaller space. China has considered building a 1228m “Bionic Tower” to house 100,000 people. The cost estimates for such a structure are in the range of $20 billion.

A Space Elevator, the type that would lift packages and passengers to geosynchronous orbit, would be about 100,000 km in height. Liftport ambitiously claims one could be built by 2018, through the use of automated robotic builders using buckytubes (buckminsterfullene) as a building material. (A buckytube is a molecule made of long chains of covalently bonded carbon atoms.) Buckytube towers would be able to easily withstand airplane impacts, or even indirect nuclear attacks.

It could take a while, but once the first orbit-reaching structure is created, there will be more to follow. The convenience of reaching space by climbing rather than launching independently will bring the cost of space travel down to affordability. The political and economic reasons for building such towers will be pursuasive to many countries and groups in the years after they become technologically feasible. We’d better get used to these towers, because we’ll be seeing a lot of them!

18 Responses »

As humanity moves into the future, our ability to control our surroundings tends to increase. Unless we blow ourselves up first, this trend is likely to continue. Eventually we shall even obtain control of processes and structures on the atomic level, through nanotechnology. Artificial Intelligence and Brain-Computer Interfacing will permit our thoughts to be instantiated as reality rapidly (within certain bounds, hopefully). The shape of the world will closely reflect our deepest desires. And what are those desires?

99% of human evolution occurred on the African savannas. Our genetically inborn preferences are those which contributed the most to survival in this context. As our technology and culture evolved, our preferences did as well - but in ways pre-established by our basic genetic template. For example, humans like flowers. That’s because flowers signified a lush environment, with ample quantities of fresh water and fruit. When humanity discovered watercolors and printing, one of the first things we created were images flowers. So we like it when we see paintings of flowers, or girls wearing dresses with floral patterns.

Human males like the idea of triumph in combat. Us males love to defeat wild animals, our rivals, and anything else that deserves smashing. Smashing makes us safe and gives us access to the best females. First we used our fists, then rocks, then throwing sticks with sharpened points. Eventually we created metalworking, and began to wield a tool often more effective than all of these - the sword. The sword is an amplified archetype in the category “tool used to achieve victory”. (The lightsaber is an example of an even more greatly amplified archetype.)

When we gain the ability to manipulate reality easily, most people will probably not choose to live within the sanitized white hallways of science fiction or the boring monoliths of The Jetsons. We will create more forests, rolling grasslands, huge gardens, splendid castles, and other things we can’t yet imagine. We’re all human, and most humans foster a romantic yearning to recreate some idealistic past. The true past was a place of disease and suffering, but we love the pleasant outlines transmitted to us through stories and our imaginations. When high control over nature is achieved, I predict that the world which shall be created will not be closely associated with “science fiction”, or the cities of the modern age, but the moderately populated, calm worlds of the past - minus the diseases and medieval torture.

And the world will remain that way until we get bored of it. Some people would get bored in hours, others weeks, but I think that most will enjoy a romantic recreation of our past for decades if not centuries. Only when the majority of us tire of this idealistic world shall something genuinely new be created. And that’s fine by me! Humanity deserves what it truly wants, not what futurists or science fiction declare is most likely to happen.

6 Responses »

An organization that has been displaying some activity lately is the Lifeboat Foundation. I’ve been asked to join their advisory board. My reply was yes, because I think it’s a good cause, the niche was bound to be filled eventually, and I’d like to be involved. The main focus of the organization, which is currently run part-time by a single individual supported by dozens of big names, is to build a space ark as an insurance policy against global technological disaster. The idea is to start with an orbital space station, then create improved versions incrementally more distant from the source of potential trouble (orbiting the Earth, then the Moon, then Mars).

My perspective on the issue is that most disasters can be avoided by simply going underground in an isolated area. Going into space seems somewhat too excessive and expensive to be realistically attainable in the short term (5-10 years). However, the organization will sponsor research in self-sustaining habitats, a technology with useful applications for underground living as well. But if it’s a disaster that a 5-stories-underground self-sustaining habitat in Antarctica can’t handle, then I doubt a space ark will help much. The primary concerns:

Nuclear war: famous studies have shown that many millions would survive.
Killer virus: extremely unlikely to wipe out literally everyone.
Grey goo: this is an implausible risk, sensationalized by clueless journalists.
Nano-dictator: if they have nanotech and you don’t, you’re finished.
Unfriendly AI: there’s nowhere to hide from a self-improving superintelligence.

How would an orbital lifeboat be useful? Perhaps in some kinds of nanotech arms races, or possibly other risks we haven’t considered yet. Certain people will work towards self-sustaining space habitats no matter what, so they might as well be organized. The Lifeboat Foundation is a wise initiative in this direction.

One Response »

This is a migration to Wordpress from Typepad. Originally I wanted to use MoveableType rather than Wordpress, but found it rather difficult to install. (Allegedly the average install time is one hour.) Wordpress, however, is free, installs in 5 minutes, and has features comparable to MoveableType. Hooray for Wordpress. All posts prior to this one are transferred from the TypePad blog.

3 Responses »

SENS stands for Strategies for Engineered Negligible Senescence, a detailed plan for reversing human aging. It is an engineering approach that seeks to slow and then halt aging processes that are the side effects of our body’s metabolic cycles. The proposal originates with Dr. Aubrey de Grey, a Cambridge biogerontologist who has appeared in CNN, the New York Times, New Scientist, Popular Science, MIT’s Technology Review, Fortune magazine, BBC News, etc. De Grey’s Methuselah Institute has raised $3M in donation committments towards the Methuselah Mouse Prize, which rewards researchers who achieve breakthroughs in substantially extending the lifespan of middle-aged laboratory mice. After reliable life extension is achieved with mice, therapies for humans would follow.

The SENS website lists the seven causes of pathogenic damage underlying aging:

1) cell depletion
2) chromosomal mutations (cancer)
3) mitochrondrial mutations
4) unwanted cells that won’t die
5) extracellular crosslinks
6) extracellular junk
7) intracellular junk

These seven sources of damage are treated as comprehensive because they were all discovered over 20 years ago, and our tools for detecting sources of pathology has improved so greatly over this time, that if there were others to be found, they would be obvious by now. De Grey proposes the following solutions which respectively correspond to the seven causes of aging:

1) Stem cells, growth factors, exercise
2) WILT (Whole-body Interdiction of Lengthening of Telomeres)
3) Allotopic expression of 13 proteins
4) Cell ablation, reprogramming
5) AGE-breaking molecules/enzymes
6) Phagocytosis; beta-breakers
7) Transgenic microbial hydrolases

De Grey proposes a 50/50 chance that within twenty to thirty years, our implementations of the above countermeasures will become sophisticated enough to lower the rate of aging to negligibility. After that point, the only threats to life which would remain are disease, war, accidents, and technological or natural disasters. Of course, success in this endeavor will require adequate funding. And the old guard biogerontologists and skeptics are coming around, bit by bit, as they realize the scientific feasibility of de Grey’s proposals.

A quick run-down of the problems and proposed solutions.

As we grow older, certain cells die without being replaced. This happens in critical organs such as the heart and brain. To fix this problem, we must encourage exercise, artificially stimulate cell growth, and apply stem cell therapy. This will ensure that cells are being reliably replaced at the same rate as they perish.

Cells need telomeres of a certain length to reproduce. As a safeguard against unconstrained cellular division (cancer), human cells have a limited quantity of telomerase (the enzyme that extends telomeres). In cancer, chromosomes mutate such that excess telomerase is produced and the cell can divide indefinitely (leading to a tumor). As a solution, we deactivate the genes that code for telomerase. No more telomere lengthening, no more cancer. The only condition is that fresh stem cells must be introduced every decade or so, because the ability of the body’s cells to divide independently is curtailed - for a worthy cause.

Mitochrondria are organelles within the cell which take oxygen and nutrients and turn them into carbon dioxide and ATP to power the cell. This is the process of breathing. Mitochrondria have their own DNA, which are susceptible to pathogenic mutations. Although mitochrondria are made up of over a thousand proteins, only 13 are manufactured within the mitochrondia itself - the rest are manufactured in the nucleus of the cell are delivered from the outside. To ensure that these 13 proteins are continuously restocked, we must modify the cellular DNA to produce these new proteins and deliver them to the mitochrondria. That way, if the mitochrondrial DNA mutates and these proteins stop being manufactured locally, it won’t matter.

As we age, certain unwanted cells start building up. Fat cells, senescent cells, and certain immune system cells. We eliminate these either through destroying them physically (by injecting something which makes only those cells commit suicide) or by programming the immune system to take them out. The latter will require some stem cell reprogramming, which has to be done anyway.

The last three categories have to do with certain types of molecular junk inside and outside the cell. Certain proteins hang around without getting recycled, and start forming unwanted bonds with each other, producing plaque. Luckily these bonds are biologically unusual so drugs have already been created which identify them and break the bonds. Other junk include lipids making up arterial plaque and amyloids, which are especially abundant in the brains of Alzheimer’s sufferers. Again, the primary approach being proposed to address this is the stimulation of the immune system to dissolve the material.

Junk also builds up inside the cell. The proposal for getting rid of this is a bit more exotic. It involves isolating enzymes within microorganisms in the soil adapted to consuming these specific proteins from the insides of cadavers. It could take work to find the right, non-toxic enzymes, but we have no reason to believe it can’t be done. Ironic that the products of death could contribute to extending the lives of the living.

So that’s it. A proposal for eliminating aging. Quite a project, but this framework gives us an abundance of starting points. And people are beginning to take it more seriously.

SENS is an effort I encourage people to give to that surpasses the humanitarian value of the vast majority of conventional charities. Following SENS are two additional efforts whose value is more difficult to explain and justify, but whose importance I consider even greater. The benefits that would flow from the success of these other two efforts are supersets of the benefits which would flow from SENS. I will cover these efforts tomorrow and the next day.

3 Responses »