Accelerating Future Lexicon:

causal functionalism:
From this dictionary of philosophy of mind: "One of the key ideas behind causal functionalism is that a physical system realizes a mental state not in virtue of the particular stuff it is made of but instead in virtue of the causal relations that parts of that system bear to each other". Causal functionalists argue that there is nothing particularly special about the building blocks of human minds, and that minds can in principle be built using different materials than proteins and neurons. In fact, a mind can theoretically be created using any sort of Turing Machine at all, including pebbles and twigs clicking into one another in an orderly manner, the entire population of China acting out a complex, interconnected dance, and so on. The essence of a mind is its information-processing routines; not the substrate these routines are implemented on. The notion that only biological neurons can implement intelligent behavior is very anthropocentric. See also uploading.

ceiling being:
A hypothetical futuristic entity approaching the physical limits of intelligence by making maximally efficient use of available matter and computing power. A potential end-state of the universe. Very speculative. See also eschatology, Alpha-Point computing, computronium, superintelligence, Omega Point Theory, Power.

Center for Responsible Nanotechnology (CRN):
Nonprofit think tank for nanotechnology (sometimes called "molecular manufacturing") policy. If nanotechnology does arrive before a Friendly Singularity, then it will be essential that peace is kept and progress towards a stable, positive society is sustained. The Center for Responsible Nanotechnology is one of the only organizations seriously pursuing this goal. With billions per year going into nanotech research, it is essential for our continued existence that we be prepared for the arrival of the molecular manufacturing revolution,and right now, we're not. Website: http://www.crnano.org. See also Foresight Institute, Grey Goo, nanotechnological arms race.

CFAI:
See Creating Friendly AI.

Chalmers, David:
David Chalmers is Professor of Philosophy and Director of the Center for Consciousness Studies at the University of Arizona. Frequently mentioned along with the notion of qualia, and the "hard problem" of conscious experience. His website is probably the best resource for the scientific study of consciousness on the web. Personal site: http://www.u.arizona.edu/~chalmers/.

cognitive science:
Field of study concerned with the brain (especially the human brain) and its functions, including high-level thought and consciousness. At the lowest level of abstraction, involves the study of neurons and neurotransmitters, proceeding up to the level of neurological modules and prefrontal cortex integrative functions. Cognitive science is a vast and complex field, and is ultimately more relevant to the study of the Singularity than human history, technology, or culture. To cognitive scientists, general intelligence is not some opaque mysterious package handed to us from on high, but a complicated machine to be analyzed and reduced (and eventually reverse-engineered). See also evolutionary psychology; Singularity, cognitive interpretation of.

collapsarity:
A possible outcome of the Singularity in which exponentially increasing intelligence is accompanied by a continuous spatial compactification of the hardware implementing that intelligence, as the intelligence makes use of ever-smaller and more energetic particles as computing elements. A collapsarity could permit the creation of a local Omega Point or aleph without a Big Crunch, possibly allowing an asymptotic quantity of computation before the final collapse. (In the tradition of Zeno's paradox, if we can keep making use of better and better computing elements as the universe gets more and more compact and energetic, then we might be able to extract a very large, possibly infinite amount of computation before the End, and therefore life and intelligence.) All of this, of course, assumes that life and intelligence can be implemented on any substrate, including exotic substrates like quarks or nucleons. This idea is known as causal functionalism. The idea of a collapsarity is advocated by Singularity analyst John Smart. Paul Hughes' term. Very speculative. See also aleph.

complex functional adaptation:
An adaptation composed of multiple, interacting components that work together to perform a function. Complicated functions do not occur in single mutations - they build up gradually, as simple implementations of the functionality allow for more and more complex improvements to take place [Yudkowsky01]. Both altruism and retaliation might be regarded as complex functional adaptations. Because they are computationally complex, we shouldn't expect arbitrary AIs to exhibit or desire these qualities. See also evolution, evolutionary biology, evolutionary psychology.

computronium:
Matter that has been transformed from its natural state into an optimized, maximally efficient computer. What constitutes "computronium" varies with the level of postulated technology. A rod logic nanocomputer is probably too primitive to qualify as computronium, since large molecular aggregates (hundreds or thousands of atoms) are used as computing elements. A more archetypal computronium would be a three-dimensional cellular automaton, which attached computational significance to each individual atom, perhaps with quantum-computing elements included [Yudkowsky01]. Contrary to popular worry, we have no particular reason to believe that self-improving intelligences or AIs would have an innate desire to convert raw materials into computronium, although it is one possible (and significant) risk. Somewhat speculative. See also aleph, nanotechnology, rod logics, Riemann Hypothesis Catastrophe.

Consensus, the:
The Consensus is the world of shared perceptions that humanity inhabits. Things in the Consensus aren't really really real, but they usually correspond tightly to reality - enough to make the rules about what you can and can't say just as strict. What distinguishes the Consensus from actual reality is that there is no a priori reason why things should be formalizable, philosophically coherent, or unambiguous [Yudkowsky01]. Non-speculative. See also objective reality, Veil of Maya.

controlled ascent:
An ethical injunction which states that self-improvement should not outpace programmer guidance, because this increases the probability of catastrophic failure of Friendliness. A consequence of this instruction would be that, if self-improvement suddenly accelerates, it may be necessary to consult the programmers before continuing, or to deliberately slow down for some period of time in order for Friendship to catch up with intelligence. For very young AIs, "controlled ascent" may be a programmatic feature triggering a save-and-stasis if a metric of self-improvement suddenly accelerates, or if the metric of self-improvement outpaces a metric of programmer guidance [Yudkowsky01]. See also Friendly AI, Singularity-ready, takeoff-safe AI.

convergent subgoal:
A subgoal that keeps popping up in generalized goal systems; subgoals that follow automatically from large classes of high-level goals. One example might be "acquire greater personal effectiveness", or "increase the intelligence I have available to accomplish goals with". If converting large amounts of local matter into computronium is a convergent subgoal (as it seems like it is), then UnFriendly AI would be a serious danger. (It doesn't seem like altruism, Friendliness, or your preferred philosophy are convergent subgoals for AIs; they're too complex structurally, as I've already mentioned several times throughout this lexicon.) Examined in more detail in part 3.2.7.1 of CFAI; the section on convergent subgoals, at http://www.singinst.org/CFAI/design/generic.html. See also Ethics in Advanced Artificial Intelligence, goal systems, paperclip AI, thermostat AI.

Cosmides & Tooby:
Leda Cosmides and John Tooby, famous evolutionary psychologists, well known for writing articles together. Along with Jerome Barkow, edited the collection of articles making up the flagship publication of the field of evolutionary psychology, "The Adapted Mind". See their Evolutionary Psychology Primer at http://www.psych.ucsb.edu/research/cep/primer.html. See also evolutionary psychology, Integrated Causal Model.

countersphexishness:
Aversion to repetitive activity. Relevant to analyses of boredom and Singularity Fun Theory. Douglas Hofstadter's term. See also sphexishness and Singularity Fun Theory.

Creating Friendly AI (CFAI):
Book-length work arguing that robustly benevolent Artificial Intelligence is possible and desirable [Yudkowsky01]. Delves deeply into design and policy implications. A publication of the Singularity Institute for Artificial Intelligence. Located at http://singinst.org/CFAI. See also 24 Definitions of Friendly AI, General Intelligence and Seed AI, Levels of Organization in General Intelligence.

Dawkins, Richard:
Famous Darwinian author who writes about science, atheism, and evolution. Professor at Oxford University. Past works include "The Selfish Gene", "The Extended Phenotype", "The Blind Watchmaker", "River Out of Eden", "Climbing Mount Improbable", and most recently, "Unweaving the Rainbow". Non-official website: http://www.world-of-dawkins.com.

de Garis, Hugo:
Associate professor of computer science at Utah State University. Pioneer of neural nets and evolvable hardware. Website at http://www.cs.usu.edu/~degaris/.

design-contingent philosophy:
The phrase "design-contingent philosophy" notes that the philosophy of a mind will be a function of 1) its initial design, 2) how acquiring new knowledge affects the behavioral output of that design (if it does at all), and 3) any revisions the mind makes to its own design, to the extent of its capability. There are no indicators that arbitrarily constructed minds tend towards structurally complex human-characteristic emotions or tendencies, such as love, selfishness, revenge, aesthetics, common sense, or your preferred philosophy. These concepts are semantic primitives to us humans because our genes gave us complex hardware support for thinking about them easily. Cognitive theorists have imagined minds with very simple goals, such as producing as many paperclips as possible [Bostrom03], minds that help others regardless of how they are treated [Yudkowsky01], minds that manage entire societies [Wright01], and so on (for various speculation, see science fiction novels). Michael Anissimov's term. See also Friendly AI, paperclip AI, thermostat AI.

design pressure:
A persistent order-generating force of some kind. Two common examples would be "evolutionary design pressure" (what is invoked in the creation of a kitten) and "intelligent design pressure" (what is invoked in the creation of a software program). In evolutionary psychology and biology, often referred to in the context of "selection pressure".

design signature:
Characteristic patterns left on an object or mind by its designer. These patterns naturally will reflect the strategy of the designer initially, but may transform over time, through external influences or self-direction. One obvious example would be the use of contraceptives by human beings. (Evolution, if it could possess feelings, would despise contraceptives.) Design signatures may be very specific, and studying the characteristic design signature of a process can give us vast insights into the form and function of the designed objects. For example, all biological studies are informed by what we know about the design signature of evolution. One aspect of the design signature of evolution on Earth seems to be that, when it creates minds at all, these minds must possess self-interested tendencies (otherwise they wouldn't be adaptive, and they couldn't remain sturdy under a range of possible mutations or environmental changes). See also evolutionary biology, evolutionary psychology.

Drexler, Dr. Eric:
Founding father of the now widely-known fields of nanotechnology and nanoscience. Vocal policy advocate on emerging technologies and their consequences for the future. Awarded a Ph.D in Molecular Nanotechnology, the first of its kind, from MIT. Co-founder and Chairman of the Foresight Institute. The Foresight Institute webpage is here: http://www.foresight.org/FI/Drexler.html.

Eclectic Pseudoplague:
The combination of tried and true killing substances, such as anthrax spores or Sarin gas, with bulk-manufactured intelligent delivery devices, as a terrorist weapon. Mitch Howe's term. See also existential risk.

EEG:
See electroencephalogram.

electroencephalogram (EEG):
A graphical record of electrical activity of the brain; produced by an electroencephalograph. See also cognitive science.

empirical regularity:
A statistical regularity that can be empirically measured, and confirmed by different observers. Empirical regularities are the fabric of order that makes our world coherent and real.

eschatology:
Study of the end of the universe, or the end of human civilization, depending on which definition you're using. Sometimes, but not always, used in the context of religion. Although the word isn't used explicitly, many Singularity analysts and other transhumanists are implicitly studying the eschatology of "exclusively human" civilization. Meanwhile, cosmologists such as Frank Tipler and Stephen Hawking are studying the eschatology of the universe itself. Several cosmological eschatologies include "The Big Crunch", "Heat Death", and "The Big Rip". Somewhat speculative.

ethical injunction:
In Friendly AI theory, a cooperative safeguard between programmers and an in-progress AI, to prevent the AI from taking actions that neither the AI nor the programmer want (such as the AI rewriting itself to be unFriendly.) Exists not as an absolute rule, but because the probability of mistaken violation is greater than the probability of correct violation [Yudkowsky01].

ethics:
The heuristics we use to choose means, given ends. An ethical principle derived from the endgoal "help others" might include "always ask people what they want before performing an action which might disturb them". An ethical principle derived from the endgoal "make as many paperclips as possible" might include "if an object seems like a potential source of metal, check it". See also metamorality, morality.

Ethics in Advanced Artificial Intelligence:
Oxford philosopher Dr. Nick Bostrom's recent paper [Bostrom03] explaining his support for accelerating the Singularity and safeguarding its integrity. Located online at http://www.nickbostrom.com/ethics/ai.html.

evolution:
The process of biological reproduction, random mutation, and natural selection that is responsible for the creation of every life form on Earth, including humans. For clarity's sake, many scientists use the word "evolution" only in the context of biology and natural selection, rather than to mean "generalized progress". See also evolutionary psychology.

evolutionary biology:
The study of how animals (including humans) evolved and which adaptive challenges they evolved to solve. A good intro can be found at http://www.talkorigins.org/faqs/faq-intro-to-biology.html. Understanding evolutionary biology is highly advisable if you want to understand the basics of Friendly AI. See also evolutionary psychology.

evolutionary psychology:
A way of looking at human psychology in the context of our evolution, most of which took place while we led hunter-gatherer lifestyles. Evolutionary psychology attempts to delineate specific cognitive modules and their function, which evolved in response to various selection pressures. The best online introduction available can be found at www.psych.ucsb.edu/research/cep/primer.html. See also complex functional adaptation, Integrated Causal Model, Massive Modularity Hypothesis.

existential risk:
A risk to the existence of humanity or life-in-general [Bostrom99]. If the abrupt and unnatural destruction of one life is bad, then the destruction of, six billion lives would be six billion times as bad, or worse. One death is a tragedy, six billion deaths is not a statistic; it is a massive tragedy. Possible existential risks include a nanotechnological arms race, a killer virus, or an UnFriendly AI. Nuclear weapons, ice ages, or asteroid impacts are far less likely to wipe humanity out entirely, although they could do a lot of damage if they occured (lower probability, though). The threat of existential risks is a big motivation for the construction of Friendly AI, or any smarter-than-human, kinder-than-human intelligence. See Dr. Nick Bostrom's original paper on existential risks at http://www.nickbostrom.com/existential/risks.html. See also UnFriendly Singularity, nanotechnological arms race.

extreme life extension:
Life extension for thousands or millions of years, as opposed to decades or centuries. In all likelihood, this would require nanotechnology and uploading, rather than mere biotechnology or conventional medicine. For more, see http://www.imminst.org or Answering Leon Kass. Somewhat speculative. See also Immortality Institute, Singularity Fun Theory.

Failure of Friendliness (FoF):
A mistake causing an otherwise Friendly AI to take one or more unFriendly actions, perhaps due to an error in supergoal content or a misapprehension about external reality. The error may or may not be recoverable [Yudkowsky01]. See also catastrophic failure of Friendliness.

femtotechnology:
Femto-scale machines and structures. A femtometer is a billionth of a meter. If they were possible, femtomachines could literally transform iron into gold or hydrogen into uranium via manipulation of the nucleons in the atom. Femtotech machines would require quarks or still smaller particles as building blocks and moving parts. If femtotech ever becomes technologically feasible, it could become practical to emulate entire civilizations using only small ensembles of atoms. (Assuming causal functionalism is true, of course.) Very speculative. See also nanotechnology.

Foresight Institute:
From their website: "Foresight Institute's goal is to guide emerging technologies to improve the human condition. Foresight focuses its efforts upon nanotechnology, the coming ability to build materials and products with atomic precision, and upon systems that will enhance knowledge exchange and critical discussion, thus improving public and private policy decisions." See http://www.foresight.org. See also Center for Responsible Nanotechnology, Drexler, Eric, Grey Goo, nanocomputing, nanotechnology, nanotechnological arms race.

Friendliness:
Intuitively: The set of actions, behaviors, and outcomes that a human would view as benevolent, rather than malevolent; nice, rather than malicious; friendly, rather than unfriendly; good, rather than evil. An AI that does what you ask ver to, as long as it doesn't hurt anyone else, or as long as it's a request to alter your own matter/space/property; an AI which doesn't cause involuntary pain, death, alteration, or violation of personal environment [Yudkowsky01]. Yudkowsky has also defined Friendliness as "...a means whereby a genuinely, willingly altruistic programming team transmits genuine, willing altruism to an AI, packaged to avoid damage in transmission, and infused in such form as to eventually become independent of the original programmers". Non-speculative. See also 24 Definitions of Friendly AI, Creating Friendly AI, Friendly AI, Friendship.

Friendly AI (FAI):
The field of study concerned with the production of human-benefiting, non-human-harming actions in Artificial Intelligence systems that have advanced to the point of making real-world plans in pursuit of goals. The term "Friendly AI" was chosen not to imply a particular internal solution, such as duplicating the human friendship instincts, but rather to embrace any set of external behaviors that a human would call "friendly". In this sense, "Friendly AI" can be used as an umbrella term for multiple design methodologies [Yudkowsky01]. Also used to refer to a completed Friendly AI itself. An AI that is Friendly not because it is being forced, but because it wants to be. "Friendly" is very different than "friendly", although the former would surely incorporate aspects of the latter. Non-speculative. See also 24 Definitions of Friendly AI, Creating Friendly AI, seed AI, Singularity.

Friendly Singularity:
A scenario in which a smarter-than-human, kinder-than-human being (perhaps a Friendly AI) achieves recursive self-improvement, attains superintelligence, and proceeds to help all humans in the ways we want to be helped. Since a superintelligence would have vast cognitive resources to tackle any problem, a Friendly Singularity might lead to the complete or near-complete elimination of involuntary death, poverty, ignorance, injury, even small annoyances. This Friendly superintelligence could assist human beings (and any sentient animals) in their own ascent to superintelligence (if they so wished), and cooperate with humans in minimizing the threat of UnFriendly AI, a nanotechnological arms race, or additional risks we aren't aware of yet. Some have argued that intelligence must necessarily be selfish, making a Friendly Singularity impossible. Yet others have argued that intelligence should be selfish, making a Friendly Singularity philosophically undesirable. Most Singularity activists, however, believe in the possibility and desirability of benevolent, self-improving transhuman intelligence. A Friendly Singularity might lead to a best-of-all-worlds scenario; an apotheosis. Explicit supporters of the Friendly Singularity idea include Dr. Nick Bostrom, Dr. Ben Goertzel, Michael Anissimov, Eliezer Yudkowsky, and several hundred others. Non-speculative. See also UnFriendly Singularity.

"Friendly" superintelligence:
An altruistic, philanthropic mind possessing level of ability (intellectual and physical) enormously in advance of that possessed by humans. In the recent paper "Ethical Issues in Advanced Artificial Intelligence", Dr. Nick Bostrom writes; "It is hard to think of any problem that a superintelligence could not either solve or at least help us solve. Disease, poverty, environmental destruction, unnecessary suffering of all kinds: these are things that a superintelligence equipped with advanced nanotechnology would be capable of eliminating. Additionally, a superintelligence could give us indefinite lifespan, either by stopping and reversing the aging process through the use of nanomedicine, or by offering us the option to upload ourselves. A superintelligence could also create opportunities for us to vastly increase our own intellectual and emotional capabilities, and it could assist us in creating a highly appealing experiential world in which we could live lives devoted to in joyful game-playing, relating to each other, experiencing, personal growth, and to living closer to our ideals". Why would a benevolent superintelligence necessarily remain benevolent? Because it would want to, and because we know of no universal force with the ability to suddenly transform a willing altruist into a selfish or otherwise unFriendly agent. A superintelligence would also have the ability to observe and manipulate its own source code, giving it further ability to preserve its own altruism under an arbitrary range of external events, including future philosophical realizations. See also Friendly Singularity.

Friendship:
The systems, subsystems, goal system content, system architecture, and other design features constituting the implementation of Friendliness [Yudkowsky01].

Friendship acquisition:
The second-order problem of Friendly AI; building a Friendly AI that can learn Friendship content [Yudkowsky01]. The cognition used to verify, modify, improve, and contradict Friendship content. (Note that "Friendship content" thus has the connotation of something that can be verified, modified, improved and contradicted without that posing an unusual problem.)

Friendship architecture:
The challenges of Friendship acquisition and Friendship structure, as distinguished from Friendship content. The part of the problem that is solved by creating specific cognitive processes, rather than by adding specific knowledge or other cognitive content. The bounded amount of complexity that is infused by design and forethought, rather than the open-ended amount of complexity that is accumulated through experience [Yudkowsky01].

Friendship content:
The zeroth-order and first-order problems of Friendly AI; correct decisions and the cognitive complexity used to make correct decisions. The complex of beliefs, memories, imagery, and concepts that is used to actually make decisions. Specific subgoal content, supergoal content, shaper content, and so on [Yudkowsky01].

Friendship structure:
The third-order problem of building a Friendly AI that wants to learn Friendliness (engage in Friendship acquisition of Friendship content). The structural problem that is unique to Friendly AI. The challenge of building a funnel through which a certain kind of complexity can be poured into the AI, such that the AI sees that pouring as desirable at every point along the way. The challenge of creating a bounded amount of Friendship complexity that can grow to handle open-ended philosophical problems [Yudkowsky01].

future-incomprehensibility:
When a smarter-than-human intelligence is created in a world composed of solely human-level intelligences, prediction of the specific nature of later events and advances (for unaugmented humans) becomes difficult because they will be driven by superintelligent forces. The only potential prediction one can make is that the actions of the first greater-than-human intelligence might somehow bear a relation to its initial goals. For example, we can guess that if Gandhi became the first superintelligence, he might choose to be altruistic, whereas if Hitler became the first superintelligence, he might choose to be authoritarian. Although there may be many things that do not change with the rise of superhuman intelligence, it's extremely difficult for us to say beforehand what they will be; we'd need to predict the precise thoughts and actions of the transhuman intelligences behind the Singularity, and there could be quite a few. The course of the future after the Singularity will depend on the values of the superintelligences directing progress in this world, which would hopefully include the freedom and survival of all sentient beings. Future-incomprehensibility wouldn't mean that our ability to understand in the world anything suddenly collapses, just that unenhanced humans might have difficulty comprehending anything but the most general aspects of superintelligent decisions or behavior. See also Friendly Singualrity, Singularity, UnFriendly Singularity.

Future Shock Level:
The quantity and extremity of futurist thought one has been exposed to [Yudkowsky98]. For example, reading about science fiction or nanotechnology would tend to increase the average person's future shock level. This idea of future shock levels is derived from Eliezer Yudkowsky's original essay, available here: http://sysopmind.com/sing/shocklevels.html. See also Shock Level Zero, Shock Level One, Shock Level Two, Shock Level Three, Shock Level Four.

Future Shock Levels:
Eliezer Yudkowsky's original essay on Future Shock Levels. Available here: http://sysopmind.com/sing/shocklevels.html. See also Shock Level Zero, Shock Level One, Shock Level Two, Shock Level Three, Shock Level Four.