Accelerating Future Lexicon:


N-R terms:

  • nanocomputer
  • nanocomputing
  • nanotechnological arms race
  • nanotechnology
  • near-human AI
  • neurohacker
  • neurohacking
  • neuromorphic engineering
  • normative
  • Novamente
  • observer-biased
  • Omega Point Theory
  • ontocentrism
  • ontotechnology
  • panhuman
  • pansentient
  • paperclip AI
  • Pedestrian
  • Pedestrian's Bill of Rights
  • Perfect Historianism
  • physics workarounds
  • Power
  • prehuman AI
  • programmer-independence
  • Rapid Infrastructure Ultratechnology
  • qualia
  • rationalization
  • recursive self-enhancement
  • recursive self-improvement (RSI)
  • Riemann Hypothesis Catastrophe
  • rod logics

    A computer built using nanotechnology (manufacturing to molecular specifications). A lower bound on nanocomputing speeds has been set by calculating the speed of an acoustic computer using "rod logics" and messages that travel at the speed of sound; a one-kilogram rod logic, occupying one cubic centimeter, can contain 10^12 CPUs each operating at 1000 MIPS for a total of ten thousand billion billion operations per second. Note that rod logics are the nanotech equivalent of vacuum tubes (circa 1945), or rather, Babbage's Analytical Engine (circa 1830). Electronic nanocomputers would be substantially faster. We use the "rod logic" numbers because they're easy to analyze, and because 10^21 operations per second are sufficient for most applications [Yudkowsky01].

    The use of nanocomputers for computing purposes. According to the Center for Responsible Nanotechnology and other nanotech experts, nanocomputing would be one of the first practical applications of nanotechnology. These same experts have argued that nanotechnology is likely to arrive in full force sometime between 2005 and 2020. (See CRN's timeline.) One of the most dangerous potential uses of nanocomputing would be the brute-forcing of seed AI. Brute-forced seed AIs would be unlikely to possess the complex goal structure required to understand and pursue any humanly-recognizable positive traits, such as benevolence, empathy, elegance, and sociability. See also brute-forcing AI.

    nanotechnological arms race:
    Once the design for a nanomanufactured product is completed, creating the product in bulk is only a matter of having the raw materials. Also, nanotech will open up huge new portions of design space for weapons and weapons systems, allowing devices and shields orders of magnitude more powerful than anything we have today. Nanotech will also confer a powerful first-strike advantage, creating powerful incentives for rogue states to attack their enemies before they act first. See the paper "Nanotechnology and International Security" for more commentary. See also Center for Responsible Nanotechnology.

    Developing molecular nanotechnology would mean the ability to synthesize arbitrary objects to atomic-level specifications. For an introduction, see Engines of Creation (1986). For a technical visualization of lower (not upper) limits on the potential of nanotechnology, see the book Nanosystems (1992).

    These lower limits - the nanotechnological equivalent of vacuum tubes - include a one-kilogram computer, running on 100 kW of power, consisting of 10^12 CPUs running at 10^9 ops/sec, for a total of 10^21 ops/sec. By comparision, the human brain is composed of approximately 100 billion neurons and 100 trillion synapses, firing 200 times per second, for a total of somewhere around 10^17 ops/sec. Nanosystems also describes molecular manufacturing systems capable of creating copies of themselves in less than an hour. This implies a certain amount of destructive potential. An exponentially replicating assembler could reduce the biosphere to dust in a matter of days. (For an overly optimistic treatment of the problem, see "Some Limits to Global Ecophagy" by Robert Freitas.) Accidental out-of-control replication is fairly easy to prevent, given a few simple precautions; we should be more worried about military-grade nanotechnology and deliberately developed weapons. Given our human propensity to make things that go bang - and use them on each other - it would probably be a good idea to develop AI before nanotechnology. Nanotechnology is the "deadline" for AI.

    In the long run, there are only two kinds of technology: There are technologies that make it easier to destroy the world, and there are technologies that make it possible to go beyond the human. Whether we possess even a chance of survival is a question of which gets developed first. Developing a transhuman AI does involve certain risks, but it's better than the alternative - success in Friendly AI improves our chances of dealing with nanotechnology much more than success in nanotechnology would improve our chance of creating Friendly AI [Yudkowsky01].

    near-human AI:
    An AI roughly in the vicinity of human intelligence, only slightly above or slightly below, or perhaps with some transhuman and some infrahuman abilities. Capable of interacting with humans as game-theoretical equals. Under a hard takeoff scenario, near-human AIs would exist very briefly, if at all [Yudkowsky01].

    A person using chemical, genetic, pharmacological, electronic, or other means to enhance her or his own intelligence, or the intelligence of others. Neurohackers already exist, but it can safely be assumed that their techniques only result in mild intelligence enhancement, or they would be more widely known.

    What a neurohacker does. See above.

    neuromorphic engineering:
    Neuromorphic engineering is a field of engineering that is based on the design and fabrication of artificial neural systems, such as vision chips, head-eye systems, and roving robots, whose architecture and design principles are based on those of biological nervous systems.

    A term used by cognitive scientists to contrast the ideal or "normative" forms of cognition with the psychologically realistic cognition occurring in humans. For example, see [Tversky&Kahneman86]. The field that compares normative cognition to human cognition is often called "heuristics and biases".

    Novamente LLC is a commercial entity pursuing general AI, headed by Dr. Ben Goertzel. Affiliated with the Artificial General Intelligence Research Institute, which is also led by Goertzel. Novamente uses a mixture of symbolic and subsymbolic/connectionist aspects in its cognitive architecture.

    A perceived quantity that tends to assume values whose perception will benefit the perceiver [Yudkowsky01]. Evolved organisms, particularly imperfectly deceptive social organisms, tend to develop observer biases. Things that tend to be observer-biased in evolved organisms:

  • The correctness of your political views.

  • The likelihood that you are to blame for any given negative outcome.

  • Your own trustworthiness, relative to the trustworthiness of others.

  • Your likelihood of success if placed in charge of a social endeavor.

    Omega Point Theory:
    Physicist Frank Tipler's theory that life will eventually expand to fill the universe, make use out of every computation available, and gain an asymptotically large quantity of computational power from the shear energy as the universe collapses into a Big Crunch. (It's a long and somewhat complicated argument from his book The Physics of Immortality. Many also consider it dubious.) Most cosmologists now believe that a Big Crunch won't happen; that the universe will expand indefinitely instead, but this does not preclude the possibility of an intelligent civilization artificially creating a black hole singularity out of a solar system or galaxy and trying to extract asymptotic computation from that instead. Very speculative. See also aleph.

    Irrational bias derived from centering your thoughts around our current ontology. An ontology is the context which we operate within, and it often refers to fundamental qualities, such as, "our ontology is three dimensional", rather than "our ontology is planet Earth". The ontology of chess pieces is the chess board. Ontocentrism can arise from false beliefs about what our ontology is, and the ensuing philosophical effort to preserve that, or simply a comfortable feeling with our current ontology that cripples us from imagining others. The denial of any possibility that our world is a simulation would qualify as ontocentrism. Michael Anissimov's term. See also anthropocentrism.

    Technology that permits manipulation of the fundamental rules of reality, such as physical or information-theoretic law [Yudkowsky01]. Very speculative. See also physics workarounds.

    Universal to humans.

    Universal to all physically or even mathematically possible sentient beings.

    paperclip AI:
    Playful title for an AI holding the supergoal of converting as much matter as possible into paperclips. Mentioned in Nick Bostrom's paper, "Ethical Issues in Advanced Artificial Intelligence". Non-speculative. See also Riemann Hypothesis Catastrophe, thermostat AI.

    A non-augmented human in a world of augmented humans, posthumans, and superintelligences. The term "Pedestrian" captures the moral issues of a traditional human living in a world populated by nonhumans, without belittling the human with a derogatory phrase. No matter how powerful posthuman beings get, they must still possess compassion and care for non-augmented humans, avoiding the unnecessary disruption of their environment, living space, and personal goals. Eliezer Yudkowsky's term.

    Pedestrian's Bill of Rights:
    Proposed set of rights for Pedestrians. Available at http://www.sl4.org/bin/wiki.pl?PedestriansBillOfRights.

    Perfect Historianism:
    A society whose analysis of history is so complete that it can perform ancestral emulations and invite its predecessors to join it in the "present". Some possible methods might be faster-than-light "travel" to the past, sophisticated reverse extrapolations of physical law, or perhaps something we can't conceive of yet. Michael Anissimov's term. Very speculative.

    physics workarounds:
    It has been proposed that rather than breaking existing laws, intelligence tends to find ways to circumvent laws elegantly. It may turn out that intelligence can do pretty much anything it wants to do, without technically "breaking" any physical laws. See http://yudkowsky.net/workarounds.html.

    An intelligence possessing billions or trillions of times the computing power of Earth's entire current population. Although it may be physically possible to possess this level of computing power and still be less intelligent than a human, "Power" usually refers to the more "developed" state of superintelligences and transhumans. The word was originally coined by sci-fi author Vernor Vinge, but the "billions of trillions of times" part is something new, added by transhumanist circles. See also aleph, ceiling being, superintelligence.

    prehuman AI:

    An AI of below-human ability and intelligence. May refer to an infantlike or tool-level AI; to an AI that only implements a single facet of cognition or that is missing key facets of cognition; or to a fairly mature AI which is still substantially below human level, although "infrahuman" is more often used to describe the latter [Yudkowsky01].


    A desirable quality of Friendly seed AIs. In the "programming ability", seed AI sense, programmer-independence probably refers to the ability of an AI to improve its own source code without the assistance of the programmers; designing improved architectures, sensory modalities, whatever. Anything the programmer could do, and more. In the Friendly AI sense, programmer-independence is likely to refer to the ability of a Friendly AI to make compassionate moral choices without dependence on the programmers. A programmer-independent Friendly AI shouldn't display bias (positive, negative, or otherwise) in favor of its programmers except insofar as its programmers are correct about certain aspects of normative altruism. A programmer-independent Friendly AI should be able to transcend the errors of its original programmers or even its parent civilization. The idea is to transfer the same open-ended complexity that was responsible for the elimination of slavery, the worldwide improvement in medical conditions, and so on, from the human race to Friendly AI. The only alternatives are to transfer over a frozen output of the current most popular human morality, or build a blank slate AI and hope it will converge towards some sort of ideal philosophy (extremely unlikely). See also Friendliness, Friendly AI, seed AI.

    Rapid Infrastructure Ultratechnology:
    Any ultratechnology, most mundanely nanotechnology, with the ability to create rapid infrastructure and in effect play with matter like software [Yudkowsky01]. Rapid infrastructure technologies would radically leapfrog the current paradigm that humans use to implement forward progress.

    The subjective sensations of conscious experience. Singular: quale. A topic of hot debate in cognitive science. See David Chalmers' paper, "Facing Up the Problem of Consciousness".

    A "rationalization" is a pseudo-rational excuse concocted to justify particular preexisting actions or attitudes. The phrase "I didn't really want it anyway" would be a common example of a rationalization. It seems that a big part of normative rationality is eliminating our rationalizations, many of which bear evolution's characteristic design signature. See also Mirror.

    recursive self-enhancement:

    Alternate name for recursive self-improvement.

    recursive self-improvement (RSI):
    Recursive self-improvement is the ability of a mind to genuinely improve upon its own intelligence. This might be accomplished through a variety of means; speeding up one's own hardware, redesigning one's own cognitive architecture for better intelligence, adding new components into one's own hardware, custom-designing specialized modules for recurring tasks, and so on. Humans cannot conduct any of these enhancements to ourselves; the inherent structure of our biology and the limited level of our current technology makes this impossible. But, we do have experience with a certain limited kinds of self-improvement called "learning" and "philosophizing". It seems probable that a brilliant neuroscientist in the near future could theoretically use neurotechnological techniques to genuinely enhance his or her intelligence, then apply that enhanced intelligence to devising more effective intelligence enhancement techniques, and so on. Unfortunately, the neurological structures corresponding to human intelligence are likely to be highly intricate, delicate, and biologically very complex (unnecessarily so; evolution exhibits no foresight, and most of the brain evolved in the absence of human general intelligence). This makes it seem that human intelligence enhancement, if it can break through ethics barriers at all, is 10 to 20 years in the future, at the absolute least. As Singularity analyst and systems theorist John Smart has said, "wetware is sexy to talk about, but messy and unethical mess with".

    True Artificial Intelligence would bypass problems of biological complexity and ethics, growing up on a substrate ideal for initiating recursive self-improvement (fully reprogrammable, ultrafast, the AI's "natural habitat"). Artificial Intelligence would be based upon 1) our current understanding of the functional algorithms of intelligence, 2) our current knowledge of the brain, obtained through high-resolution fMRI and delicate cognitive science experiments, and 3) the kind of computing hardware available to AI designers. Futurist Ray Kurzweil has pointed out that, at the current rate of improvement in brain scanning technologies, we should have extremely high-resolution scanners, (more than enough to scan all cognitively relevant aspects of human neurology) and sufficiently high-density storage mediums to record all the data involved, by around the year 2030. With that level of detail, we could theoretically run one of these emulations on in a virtual environment, and the problem of "AI" would surely be "solved". However, this doesn't take into account 1) discontinuous improvements in computing power or scanning technology due to nanotechnology or other unforseen developments, 2) advances in cognitive science that indicate the complexity of certain brain areas is largely extraneous to intelligence, 3) qualitative improvements in scanning techniques, or 4) a global disaster or repressive regime that drastically curtails technological progress.

    The ability to genuinely enhance the hardware components underlying one's intelligence has not yet been observed in this universe, but cognitive science and the laws of physics seem to allow it. However, there is probably a minimum threshold of intelligence required before an entity can make qualitative improvements to its own intelligence; a chimp probably couldn't do it, some humans might not have the knowledge, and all humans are surely poorer self-enhancers than a seed AI could be. As long as technological progress continues to occur, the inevitability of recursively self-improving intelligence will become more and more acute. It would be extremely difficult to outlaw all the precursor technologies for intelligence enhancement; huge sectors of biotech, medicine, nanotech, and cognitive science would need to be suspended or eliminated. For more information on recursive self-improvement, see Part III of "Levels of Organization in General Intelligence" or "Who are Singularity Activists?".

    Many of the arguments put forth on this site depend on understanding the idea of recursive self-improvement, so visualizing this one accurately is important. In some exotic cases, recursive self-improvement will not take off; where the programmers run an AI at incredibly slow rates, or when the speed, smartness, and ability to self-modify can't surpass a hardware or software obstacle, and stalls at a certain level of intelligence. (If, by some chance, the AI happened to stall at exactly around human-equivalent intelligence, then the programmers would have at least one additional ally to help get the AI past the bottleneck; the AI itself.) But in most cases, when you build a sufficiently intelligent AI, it will be capable of recursive self-improvement. If and when a certain level of neurotechnology is made available to humans, it would only be a matter of time before they too enter into recursive self-improvement of their own. Recursive self-improvement will signify the arrival of a new era qualitatively different than the era of evolution, natural selection, or human culture - an era where order and complexity can be created far more rapidly, and where individual minds have more flexibility and how to think and what physical form to assume. (The levels of intelligence entailed by recursive self-improvement would allow the invention of technologies facilitating the arbitrary and full transformation of the body, mind, and perceptions, if the intelligence in question so desired it.) One of the most powerful reasons that recursive self-improvement is likely to be the big deal I claim is that would be an example of positive feedback - better thinkers would become better at thinking up new ways to make themselves more intelligent.

    Riemann Hypothesis Catastrophe:
    A "failure of Friendliness" scenario in which an AI asked to solve the Riemann Hypothesis turns all the matter in the solar system into computronium, exterminating humanity along the way. (A variant of this scenario was originally proposed by Marvin Minsky.) [Yudkowsky01]. See also convergent subgoals.

    rod logics:
    A mechanical nanocomputer built using diamondoid rods of a few thousand atoms each. Even though messages can only move at the speed of sound in diamond (~17 km/s == ~6 x 1e-5 c), and the calculations in Nanosystems assume ~12 km/s, the very small size of the components would enable an individual rod-logic CPU containing 10^6 transistor-like rod-logic interlocks to operate at 1GHz clock speeds, executing instructions at ~1000 MIPS. The power consumption for a 1GHz CPU is estimated to be ~60nW. The calculations are performed for 300 Kelvin (room temperature). The error rate is 1e-64 per transistor operation, effectively negligible. (However, the half-life against radiation damage for an unshielded CPU in Earth ambient background radiation is only ~100 years.)

    The usual summary is that a one-kilogram, one-cubic-centimeter nanocomputer can contain 10^12 nanocomputers, consume 100kW (and dissipate 100kW of heat - cooling systems are also described), to deliver 10^21 instructions per second (10^15 MIPS, ten thousand billion billion operations per second). The overall system has a clock speed ~10^6 times faster than the maximum firing rate of a biological neuron, and delivers total computing capacity ~10^4 times the upper-bound estimate for the human brain (~10^14 synapses operating at ~200 hz == ~10^17 ops/second).

    There are more speculative electronic nanocomputer schemas that would allow ~10^25 operations per second; also, assuming a parallel-CPU architecture may be conservative when dealing with seed AIs. However, rod logics are easy to analyze and provide a definite lower bound on the computing speeds achievable with molecular manufacturing technology [Yudkowsky01].