Terms and Concepts

Here's a collection of terms and concepts I've learned in the course of my research so far. I hope you find some of them unfamiliar and/or interesting!

Antaḥkaraṇa #

Antaḥkaraṇa is a concept in Hindu philosophy referring to the totality of the mind. It also refers to the four functions of the mind, namely the manas (the mind or lower mind), buddhi (the intellect or higher mind), chitta (memory, or, consciousness), and ahamkara (ego, or, I-maker).

Antisyzygy #

An antisyzygy is a joining together of opposites.

Apotheosis #

Apotheosis is the glorification of a subject to divine levels.

Arithmetic coding #

Arithmetic coding (AC) is a form of entropy encoding used in lossless data compression. Normally, a string of characters is represented using a fixed number of bits per character, as in the ASCII code. When a string is converted to arithmetic encoding, frequently used characters will be stored with fewer bits and not-so-frequently occurring characters will be stored with more bits, resulting in fewer bits used in total.

Assembly theory #

Assembly theory is a framework developed to quantify the complexity of molecules and objects by assessing the minimal number of steps required to assemble them from fundamental building blocks.

Bayesian neural network #

The Bayesian neural network (BNN) model is an extension of a traditional neural network model. Each weight is a distribution rather than a single number.

Biofeedback #

Biofeedback is the technique of gaining greater awareness of many physiological functions of one's own body by using electronic or other instruments, and with a goal of being able to manipulate the body's systems at will.

Brainwave entrainment #

Brainwave entrainment refers to the observation that brainwaves will naturally synchronize to the rhythm of periodic external stimuli, such as flickering lights, speech, music, or tactile stimuli.

Centrosymmetry #

Centrosymmetry is the property of an object being identical to its mirror image.

Commensurability #

Commensurability is a concept in the philosophy of science. Scientific theories are said to be "commensurable" if scientists can discuss the theories using a shared nomenclature that allows direct comparison of them to determine which one is more valid or useful. On the other hand, theories are "incommensurable" if they are embedded in starkly contrasting conceptual frameworks whose languages do not overlap sufficiently to permit scientists to directly compare the theories.

Conatus #

Conatus is the idea that all objects have an inherent tendency to persist in their existing state, and that this tendency is the source of all action.

Cormoid #

Cormoid is an adjective meaning "similar to a corm (a type of underground storage organ in some plants)".

Critical path method #

The critical path method (CPM) is an algorithm for scheduling a set of project activities. A critical path is determined by identifying the longest stretch of dependent activities and measuring the time required to complete them from start to finish.

Cryptomnesia #

Cryptomnesia occurs when a forgotten memory returns without its being recognized as such by the subject, who believes it is something new and original.

Cunningham's Law #

Cunningham's Law is an adage that states: "the best way to get the right answer on the internet is not to ask a question; it's to post the wrong answer."

CVE #

The Common Vulnerabilities and Exposures (CVE) system provides a reference method for publicly known information-security vulnerabilities and exposures.

Defender's dilemma #

The defender's dilemma is a concept in cybersecurity that describes the idea that defenders are at a disadvantage when protecting systems from cyberattacks. The dilemma is based on the idea that defenders must be right 100% of the time, while attackers only need to be right once.

Desultory #

Desultory is an adjective meaning "lacking purpose or direction".

Discovery system #

A discovery system is an artificial intelligence system that attempts to discover new scientific concepts or laws.

Displacement activities #

Displacement activities occur when an animal experiences high motivation for two or more conflicting behaviours: the resulting displacement activity is usually unrelated to the competing motivations.

Double bind #

A double bind is a dilemma in communication in which an individual receives two or more mutually conflicting messages. A double bind generally includes different levels of abstraction in the order of messages.

Dutch book #

A Dutch book is a sequence of bets that guarantees a profit for the bookmaker regardless of the outcome of the events being bet on.

Entity-component-system #

An entity-component-system (ECS) is a design pattern used in game development to manage and organize game objects and their components.

Ergodicity #

In mathematics, ergodicity expresses the idea that a point of a moving system, either a dynamical system or a stochastic process, will eventually visit all parts of the space that the system moves in, in a uniform and random sense.

Error theory #

Error theory in moral philosophy is a position characterized by its commitment to two propositions: (i) all moral claims are false and (ii) we have reason to believe that all moral claims are false.

Ex-ante #

Ex-ante is a Latin phrase meaning "before the event".

Exegesis #

Exegesis is the act of interpreting a text, especially a religious or philosophical text.

Exploration-exploitation dilemma #

The exploration-exploitation dilemma is the balancing act between two opposing strategies: exploitation is choosing the best option based on current knowledge of the system (which may be incomplete or misleading), and exploration is trying out new options that may lead to better outcomes in the future.

Face-to-face relation #

The face-to-face relation is a concept from Emmanuel Lévinas that proposes that people are ethically responsible to one-another in a face-to-face encounter.

Feedforward neural network #

A feedforward neural network (FFNN) is a type of artificial neural network where the connections between the nodes do not form a cycle.

Fuzzy hashing #

Fuzzy hashing is a type of hashing algorithm in which two similar inputs will generate two similar hash values. This property is the exact opposite of the avalanche effect desired in cryptographic hash functions.

Gadfly #

A gadfly is a person who interferes with the status quo of a society or community by posing novel, potentially upsetting questions, usually directed at authorities.

German tank problem #

The German tank problem consists of estimating the maximum of a discrete uniform distribution from sampling without replacement. In simple terms, suppose there exists an unknown number of items which are sequentially numbered from 1 to N. A random sample of these items is taken and their sequence numbers observed; the problem is to estimate N from these observed numbers.

Good regulator #

The good regulator is a theorem in cybernetics stating that "every good regulator of a system must be a model of that system".

Grokking #

In machine learning, grokking is a transition to generalization that occurs many training iterations after the interpolation threshold, after many iterations of seemingly little progress, as opposed to the usual process where generalization occurs slowly and progressively once the interpolation threshold has been reached.

Hagiography #

A hagiography is a biography of a saint or other religious figure.

Homo reciprocans #

Homo reciprocans is the concept in some economic theories of humans as cooperative actors who are motivated by improving their environment through positive reciprocity (rewarding other individuals) or negative reciprocity (punishing other individuals), even in situations without foreseeable benefit for themselves.

Horizontal gene transfer #

Horizontal gene transfer (HGT) is the movement of genetic material between organisms other than by the ("vertical") transmission of DNA from parent to offspring.

Ideological Turing Test #

An Ideological Turing Test is a method for determining whether an ideological partisan correctly understands the arguments of their adversaries, by asking them to answer questions or write an essay from the opposite perspective.

IID task #

In machine learning, an IID task is a task that uses independent and identically distributed (IID) data. The IID assumption states that the generation of each data point in a dataset is independent of the generation of other data points.

Inductive bias #

The inductive bias of a learning algorithm is the set of assumptions that the learner uses to predict outputs of given inputs that it has not encountered. Inductive bias is anything which makes the algorithm learn one pattern instead of another pattern given the same inputs.

In-context learning #

In-context learning (ICL) is a type of machine learning where a language model is given a small set of examples and then asked to make predictions on a new example.

Interpolation #

Interpolation is the process of estimating the value of a function between two known points. This is in contrast to extrapolation, which is the process of estimating the value of a function outside of the range of known points.

Inveigle #

To inveigle someone is to persuade them to do something by means of deception or flattery.

Kludge #

A kludge is a system, especially a computer system, that is constituted of poorly matched elements or of elements originally intended for other applications.

Light cone #

A light cone is the path that a flash of light, emanating from a single event and traveling in all directions, would take through spacetime. When someone refers to "the light cone", they're usually talking about the future light cone, which represents all the events in spacetime that they could potentially influence.

Lindy effect #

The Lindy effect is a theorized phenomenon by which the future life expectancy of some non-perishable things, like a technology or an idea, is proportional to their current age; that is, the longer a period something has survived to exist or be used in the present, the longer its remaining life expectancy.

LoRA #

Low-Rank Adaptation (LoRA) is a technique for efficiently fine-tuning a pre-trained language model.

Martingale #

In probability theory, a martingale is a sequence of random variables for which, at a particular time, the conditional expectation of the next value in the sequence is equal to the present value, regardless of all prior values.

Micromort #

A micromort is a unit of risk defined as a one-in-a-million chance of death.

Minimum Description Length #

Minimum Description Length (MDL) is a model selection principle where the shortest description of the data is the best model. MDL methods learn through a data compression perspective and are sometimes described as mathematical applications of Occam's razor.

Mode collapse #

In machine learning, mode collapse is a phenomenon where a generative model learns to produce only a small range of outputs, rather than a diverse set of outputs.

Monopolylogue #

A monopolylogue is a form of entertainment in which one actor plays many characters.

Neoterism #

A neoterism is an advance in language; for example, a new word or phrase.

No Free Lunch Theorem #

The No Free Lunch Theorem states that there is no single optimization algorithm that consistently outperforms all others across all possible problems; essentially, if an algorithm performs exceptionally well on one type of problem, it must necessarily perform poorly on another type of problem.

Omohundro drive #

An Omohundro drive is an instrumental goal that an AI system might naturally develop regardless of its primary objectives, such as self-preservation and resource acquisition.

Oneiric #

Oneiric is an adjective meaning "related to dreams".

Onomastics #

Onomastics is the study of the origin, development, and meaning of proper names.

Open Individualism #

Open Individualism is the view that there exists only one numerically identical subject, who is everyone at all times, in the past, present and future. It is contrasted with "Empty individualism", the view that personal identities correspond to a fixed pattern that instantaneously disappears with the passage of time, and "Closed individualism", the view that personal identities are particular to subjects and yet survive over time.

Pessimistic induction #

The pessimistic induction, or pessimistic meta-induction, is the argument that if past successful and accepted scientific theories were found to be false, we have no reason to believe that our currently successful theories are approximately true.

Phase space #

The phase space of a physical system is the set of all possible physical states of the system when described by a given parameterization. Each possible state corresponds uniquely to a point in the phase space. For mechanical systems, the phase space usually consists of all possible values of the position and momentum parameters.

Principal-agent problem #

The principal-agent problem is a conflict of interest that occurs when an agent acts on behalf of a principal, but the two parties have misaligned incentives or unequal information. The agent may act in their own interests, avoiding responsibilities or making poor decisions that harm the principal.

Prosaic AGI #

Prosaic AGI is AGI that doesn’t reveal any fundamentally new ideas about the nature of intelligence or turn up any "unknown unknowns," as formulated by Paul Christiano.

Quadratic voting #

Quadratic voting is a voting system that works by allowing users to "pay" for additional votes on a given outcome to express their support for given issues more strongly. Quadratic funding is funding decided via quadratic voting.

Saccadic masking #

Saccadic masking is the phenomenon in visual perception where the brain selectively blocks visual processing during eye movements in such a way that neither the motion of the eye nor the gap in visual perception is noticeable to the viewer.

SAT problem #

The SAT problem is the problem of determining whether a given Boolean formula is satisfiable; that is, whether there exists an assignment of truth values to the variables that makes the formula true.

Scalable oversight #

Scalable oversight is the problem of how to oversee AI systems when they surpass human abilities.

Scope neglect #

Scope neglect is a cognitive bias that occurs when the valuation of a problem is not valued with a multiplicative relationship to its size.

Semiotic measure #

Semiotic measure is a measure of the prominence of an idea or entity in a sea of information. The higher the semiotic measure of an idea, the more influence it has over models trained on a corpus.

Sempiternal #

Sempiternal is an adjective meaning "lasting forever".

String predicate #

String predicate is a function that takes a string as input and returns a boolean (true/false) value based on whether the string meets certain criteria.

Supererogation #

Supererogation is the performance of more than is asked for or required.

Thalweg #

The thalweg is the line of steepest descent in a river.

Theodicy #

In the philosophy of religion, a theodicy is an argument that attempts to resolve the problem of evil that arises when all power and all goodness are simultaneously ascribed to God.

Time-binding #

Time-binding is the activity of transmitting experience from one generation to another. Alfred Korzybski, inventor of the term, described it as an activity that is unique to humans among all animals.

Total functional programming #

Total functional programming is a programming paradigm that restricts the range of programs to those that are provably terminating.

Trial by ordeal #

Trial by ordeal was an ancient judicial practice by which the guilt or innocence of the accused was determined by subjecting them to a painful experience, based on the premise that God would help the innocent by performing a miracle on their behalf.

Universal Darwinism #

Universal Darwinism is the extension of the theory of Darwinism beyond its original domain of biological evolution on Earth. Universal Darwinism aims to formulate a generalized version of the mechanisms of variation, selection and heredity, so that they can apply to a wide variety of other domains.

Value of information #

Value of information (VOI) is the amount a decision maker would be willing to pay for information prior to making a decision.

Verboten #

Verboten is an adjective meaning "forbidden, especially by an authority".

Vinge's Law #

Vinge’s Law, as rephrased by Eliezer Yudkowsky, states: "Characters cannot be significantly smarter than their authors. You can’t have a realistic character that’s too much smarter than the author, because to really know how a character like that would think, you’d have to be that smart yourself."

Weird machine #

In computer security, a weird machine is a computational artifact where additional code execution can happen outside the original specification of the program. When the system is somehow moved into a state that "makes no sense" when viewed from the perspective of the intended finite-state machine, the software will keep transforming the broken state into new broken states, triggered by further user input.

Wild type #

The wild type is the phenotype of the typical form of a species as it occurs in nature.