What Is Artificial General Intelligence?

Artificial General Intelligence (AGI) is defined as the software representation of generalized human cognitive capacities that enables the AGI system to solve problems when presented with new tasks. 

In other words, it's AI's capacity to learn similarly to humans.

Strong AI, full AI, and general intelligent action are some names for it. 

The phrase "strong AI," however, is only used in few academic publications to refer to computer systems that are sentient or aware. 

These definitions may change since specialists from many disciplines see human intelligence from various angles. 

For instance, computer scientists often characterize human intelligence as the capacity to accomplish objectives. 

On the other hand, general intelligence is defined by psychologists in terms of survival or adaptation.

Weak or narrow AI, in contrast to strong AI, is made up of programs created to address a single issue and lacks awareness since it is not meant to have broad cognitive capacities. 

Autonomous cars and IBM's Watson supercomputer are two examples. 

Nevertheless, AGI is defined in computer science as an intelligent system having full or comprehensive knowledge as well as cognitive computing skills.

As of right now, there are no real AGI systems; they are still the stuff of science fiction. 

The long-term objective of these systems is to perform as well as humans do. 

However, due to AGI's superior capacity to acquire and analyze massive amounts of data at a far faster rate than the human mind, it may be possible for AGI to be more intelligent than humans.

Artificial intelligence (AI) is now capable of carrying out a wide range of functions, including providing tailored suggestions based on prior web searches. 

Additionally, it can recognize various items for autonomous cars to avoid, recognize malignant cells during medical inspections, and serve as the brain of home automation. 

Additionally, it may be utilized to find possibly habitable planets, act as intelligent assistants, be in charge of security, and more.

Naturally, AGI seems to far beyond such capacities, and some scientists are concerned this may result in a dystopian future. 

Elon Musk said that sentient AI would be more hazardous than nuclear war, while Stephen Hawking advised against its creation because it would see humanity as a possible threat and act accordingly.

Despite concerns, most scientists agree that genuine AGI is decades or perhaps centuries away from being developed and must first meet a number of requirements (which are always changing) in order to be achieved. 

These include the capacity for logic, tact, puzzle-solving, and making decisions in the face of ambiguity. 

Additionally, it must be able to plan, learn, and communicate in natural language, as well as represent information, including common sense. 

AGI must also have the capacity to detect (hear, see, etc.) and output the ability to act, such as moving items and switching places to explore. 

How far along are we in the process of developing artificial general intelligence, and who is involved?

In accordance with a 2020 study from the Global Catastrophic Risk Institute (GCRI), academic institutions, businesses, and different governmental agencies are presently working on 72 recognized AGI R&D projects. 

According to the poll, projects nowadays are often smaller, more geographically diversified, less open-source, more focused on humanitarian aims than academic ones, and more centered in private firms than projects in 2017. 

The comparison also reveals a decline in projects with academic affiliations, an increase in projects sponsored by corporations, a rise in projects with a humanitarian emphasis, a decline in programs with ties to the military, and a decline in US-based initiatives.

In AGI R&D, particularly military initiatives that are solely focused on fundamental research, governments and organizations have very little roles to play. 

However, recent programs seem to be more varied and are classified using three criteria, including business projects that are engaged in AGI safety and have humanistic end objectives. 

Additionally, it covers tiny private enterprises with a variety of objectives including academic programs that do not concern themselves with AGI safety but rather the progress of knowledge.

One of the most well-known organizations working on AGI is Carnegie Mellon University, which has a project called ACT-R that aims to create a generic cognitive architecture based on the basic cognitive and perceptual functions that support the human mind. 

The project may be thought of as a method of describing how the brain is structured such that different processing modules can result in cognition.

Another pioneering organization testing the limits of AGI is Microsoft Research AI, which has carried out a number of research initiatives, including developing a data set to counter prejudice for machine-learning models. 

The business is also investigating ways to advance moral AI, create a responsible AI standard, and create AI strategies and evaluations to create a framework that emphasizes the advancement of mankind.

The person behind the well-known video game franchises Commander Keen and Doom has launched yet another intriguing endeavor. 

Keen Technologies, John Carmack's most recent business, is an AGI development company that has already raised $20 million in funding from former GitHub CEO Nat Friedman and Cue founder Daniel Gross. 

Carmack is one of the AGI optimists who believes that it would ultimately help mankind and result in the development of an AI mind that acts like a human, which might be used as a universal remote worker.

So what does AGI's future hold? 

The majority of specialists are doubtful that AGI will ever be developed, and others believe that the urge to even develop artificial intelligence comparable to humans will eventually go away. 

Others are working to develop it so that everyone will benefit.

Nevertheless, the creation of AGI is still in the planning stages, and in the next decades, little progress is anticipated. 

Nevertheless, throughout history, scientists have debated whether developing technologies with the potential to change people's lives will benefit society as a whole or endanger it. 

This proposal was considered before to the invention of the vehicle, during the development of AC electricity, and when the atomic bomb was still only a theory.

~ Jai Krishna Ponnappan

Find Jai on Twitter | LinkedIn | Instagram

You may also want to read more about Artificial Intelligence here.

Be sure to refer to the complete & active AI Terms Glossary here.

Artificial Intelligence - The General Problem Solver Software.

 

To arrive at a solution, General Problem Solver is a software for a problem-solving method that employs means-ends analysis and planning.

The software was created in such a way that the problem-solving process is separated from information unique to the situation at hand, allowing it to be used to a wide range of issues.

The software, which was first created by Allen Newell and Herbert Simon in 1957, took over a decade to complete.

George W. Ernst, a graduate student at Newell, wrote the latest edition while doing research for his dissertation in 1966.

The General Problem Solver arose from Newell and Simon's work on the Logic Theorist, another problem-solving tool.

The duo likened Logic Theorist's problem-solving approach to that of people solving comparable issues after inventing it.

They discovered that the logic theorist's method varied significantly from that of humans.

Newell and Simon developed General Problem Solver using the knowledge on human problem solving obtained from these experiments, hoping that their artificial intelligence work would contribute to a better understanding of human cognitive processes.

They discovered that human problem-solvers could look at the intended outcome and, using both backward and forward thinking, decide actions they might take to get closer to that outcome, resulting in the development of a solution.

The General Problem Solver, which Newell and Simon felt was not just reflective of artificial intelligence but also a theory of human cognition, included this mechanism.

To solve issues, General Problem Solver uses two heuristic techniques: 

1. means-ends analysis and 
2. planning.

As an example of means-ends analysis in action, consider the following: 

• If a person coveted a certain book, they would want to be in possession of it.
• The book is currently kept by the library, and they do not have it in their possession.
• The individual has the ability to eliminate the gap between their existing and ideal states.
• They may do so by borrowing the book from the library, and they have other alternatives for getting there, including driving.
• If the book has been checked out by another customer, however, there are possibilities for obtaining it elsewhere.
• To buy it, the consumer may go to a bookshop or order it online.
• The individual must next consider the many possibilities open to them.
• And so on.

The individual is aware of a number of pertinent activities they may do, and if they select the right ones and carry them out in the right sequence, they will be able to receive the book.

Means ends analysis in action is the person who chooses and implements suitable activities.

The programmer sets up the issue as a starting state and a state to be attained when using means-ends analysis with General Problem Solver.

The difference between these two states is calculated using the General Problem Solver (called objects).

• Operators that lessen the difference between the two states must also be coded into the General Problem Solver.

• It picks and implements an operator to address the issue, then assesses if the operation has got it closer to its objective or ideal state.

• If that's the case, it'll go on to the next operator.

• If it doesn't work, it may go back and try another operator.

• The difference between the original state and the target state is decreased to zero by applying operators.

• The capacity to plan was also held by General Problem Solver.

General Problem Solver might sketch a solution to the problem by removing the specifics of the operators and the difference between the starting and desired states.

After a broad solution had been defined, the specifics could be reinserted into the issue, and the subproblems formed by these details could be addressed within the solution guide lines produced during the outlining step.

Defining an issue and operators to program the General Problem Solver was a time-consuming task for programmers.

It also meant that, as a theory of human cognition or an example of artificial intelligence, General Problem Solver took for granted the very actions that, in part, represent intelligence, namely the acts of defining a problem and selecting relevant actions (or operations) from an infinite number of possible actions in order to solve the problem.

In the mid-1960s, Ernst continued to work on General Problem Solver.

He wasn't interested in human problem-solving procedures; instead, he wanted to discover a way to broaden the scope of General Problem Solver so that it could solve issues outside of the logic domain.

In his version of General Problem Solver, the intended state or object was expressed as a set of constraints rather than an explicit specification.

Ernst also altered the form of the operators such that the output of an operator may be written as a function of the starting state or object (the input).

His updated General Problem Solver was only somewhat successful in solving problems.

Even on basic situations, it often ran out of memory.

"We do believe that this specific aggregation of IPL-Vcode should be set to rest, as having done its bit in furthering our knowledge of the mechanics of intelligence," Ernst and Newell proclaimed in the foreword of their 1969 book GPS: A Case Study in Generality and Problem Solving(Ernst and Newell 1969, vii).

Artificial Intelligence Problem Solving

The AI reflex agent converts states into actions. 

When these agents fail to function in an environment where the state of mapping is too vast for the agent to handle, the stated issue is resolved and passed to a problem-solving domain, which divides the huge stored problem into smaller storage areas and resolves them one by one. 

The targeted objectives will be the final integrated action.

Different sorts of issue-solving agents are created and used at an atomic level without any internal state observable with a problem-solving algorithm based on the problem and their working area. 

By describing issues and many solutions, the problem-solving agent executes exactly. 

So we may say that issue solving is a subset of artificial intelligence that includes a variety of problem-solving approaches such as tree, B-tree, and heuristic algorithms.

A problem-solving agent is also known as a goal-oriented agent since it is constantly focused on achieving the desired outcomes.

AI problem-solving steps: 

The nature of people and their behaviors are intimately related to the AI dilemma. 

To solve a problem, we require a set of discrete steps, which makes human labor simple. These are the actions that must be taken to fix a problem:

• Goal formulation is the first and most basic stage in addressing a problem. 

It arranges discrete stages to establish a target/goals that need some action in order to be achieved. 

AI agents are now used to formulate the aim. 

One of the most important processes in issue resolution is problem formulation, which determines what actions should be followed to reach the specified objective. 

This essential aspect of AI relies on a software agent, which consists of the following components to construct the issue. 

Components needed to formulate the problem: 

This state necessitates a beginning state for the challenge, which directs the AI agent toward a certain objective. 

In this scenario, new methods also use a particular class to solve the issue area. 

Action: In this step of issue formulation, all feasible actions are performed using a function with a specified class obtained from the starting state.

Transition: In this step of issue formulation, the actual action performed by the previous action stage is combined with the final stage to be sent to the next stage.

Objective test: This step assesses if the integrated transition model accomplished the given goal or not; if it did, halt the action and go to the next stage to estimate the cost of achieving the goal.

Path costing is a component of problem-solving that assigns a numerical value to the expense of achieving the objective. 

It necessitates the purchase of all gear, software, and human labor.

General Problem Solver Overview

In theory, GPS may solve any issue that can be described as a collection of well-formed formulas (WFFs) or Horn clauses that create a directed graph with one or more sources (that is, axioms) and sinks (that is, desired conclusions). 

Predicate logic proofs and Euclidean geometry problem spaces are two primary examples of the domain in which GPS may be used. 

It was based on the theoretical work on logic machines by Simon and Newell. 

GPS was the first computer software to segregate its issue knowledge (expressed as input data) from its problem-solving method (a generic solver engine). 

IPL, a third-order programming language, was used to build GPS. 

While GPS was able to tackle small issues like the Hanoi Towers that could be adequately described, it was unable to handle any real-world problems since search was quickly lost in the combinatorial explosion. 

Alternatively, the number of "walks" across the inferential digraph became computationally prohibitive. 

(In fact, even a simple state space search like the Towers of Hanoi may become computationally infeasible, however smart pruning of the state space can be accomplished using basic AI methods like A* and IDA*.)

In order to solve issues, the user identified objects and procedures that might be performed on them, and GPS created heuristics via means-ends analysis. 

• It concentrated on the available processes, determining which inputs and outputs were acceptable. 

• It then established sub goals in order to move closer to the ultimate objective.

• The GPS concept ultimately developed into the Soar artificial intelligence framework.

~ Jai Krishna Ponnappan

You may also want to read more about Artificial Intelligence here.

See also: 

Expert Systems; Simon, Herbert A.

Frequently Asked Questions:

In Artificial Intelligence, what is a generic problem solver?

Herbert Simon, J.C. Shaw, and Allen Newell suggested the General Problem Solver (GPS) as an AI software. It was the first useful computer software in the field of artificial intelligence. It was intended to function as a global problem-solving machine.

What is the procedure for using the generic issue solver?

In order to solve issues, the user identified objects and procedures that might be performed on them, and GPS created heuristics via means-ends analysis. It concentrated on the available processes, determining which inputs and outputs were acceptable.

What exactly did the General Problem Solver accomplish?

The General Problem Solver (GPS) was their next effort, which debuted in 1957. GPS would use heuristic approaches (modifiable "rules of thumb") repeatedly to a problem and then undertake a "means-ends" analysis at each stage to see whether it was getting closer to the intended answer.

What are the three key domain universal issue-solving heuristics that Newell and Simon's general problem solver incorporated in 1972?

According to Newell and Simon (1972), every issue has a problem space that is described by three components: 1) the issue's beginning state; 2) a collection of operators for transforming a problem state; 3) a test to see whether a problem state is a solution.

What is heuristic search and how does it work?

Heuristic search is a kind of strategy for finding the best solution to a problem by searching a solution space. The heuristic here use some mechanism for searching the solution space while determining where the solution is most likely to be found and concentrating the search on that region.

What are the elements of a broad issue?

The issue itself, articulated clearly and with sufficient context to explain why it is significant; the way of fixing the problem, frequently presented as a claim or a working thesis; and the goal, declaration of objective, and scope of the paper the writer is writing.

What are the stages of a basic development process employing a problem-solving approach?

Problem-Solving Process in 8 Steps:

Step 1: Identify the issue. What exactly is the issue?

Step 2: Identify the issue.

Step 3: Establish the objectives.

Step 4: Determine the problem's root cause.

Step 5: Make a plan of action.

Step 6: Put your plan into action.

Step 7: Assess the Outcomes

Step 8: Always strive to improve.

 

What's the difference between heuristic and algorithmic problem-solving?

A step-by-step technique for addressing a given issue in a limited number of steps is known as an algorithm. Given the same parameters, an algorithm's outcome (output) is predictable and repeatable (input). A heuristic is an informed assumption that serves as a starting point for further investigation.

What makes algorithms superior than heuristics?

Heuristics entail using a learning and discovery strategy to obtain a solution, while an algorithm is a clearly defined set of instructions for solving a problem. Use an algorithm if you know how to solve an issue.

References and Further Reading:

Barr, Avron, and Edward Feigenbaum, eds. 1981. The Handbook of Artificial Intelligence, vol. 1, 113–18. Stanford, CA: HeurisTech Press.

Ernst, George W., and Allen Newell. 1969. GPS: A Case Study in Generality and Problem Solving. New York: Academic Press.

Newell, Allen, J. C. Shaw, and Herbert A. Simon. 1960. “Report on a General Problem Solving Program.” In Proceedings of the International Conference on Information Processing (June 15–20, 1959), 256–64. Paris: UNESCO.

Simon, Herbert A. 1991. Models of My Life. New York: Basic Books.

Simon, Herbert A., and Allen Newell. 1961. “Computer Simulation of Human Thinking and Problem Solving.” Datamation 7, part 1 (June): 18–20, and part 2 (July): 35–37

Artificial Intelligence - General and Narrow Categories Of AI.

There are two types of artificial intelligence: general (or powerful or complete) and narrow (or limited) (or weak or specialized).

In the actual world, general AI, such as that seen in science fiction, does not yet exist.

Machines with global intelligence would be capable of completing every intellectual endeavor that humans are capable of.

This sort of system would also seem to think in abstract terms, establish connections, and communicate innovative ideas in the same manner that people do, displaying the ability to think abstractly and solve problems.

Such a computer would be capable of thinking, planning, and recalling information from the past.

While the aim of general AI has yet to be achieved, there are more and more instances of narrow AI.

These are machines that perform at human (or even superhuman) levels on certain tasks.

Computers that have learnt to play complicated games have abilities, techniques, and behaviors that are comparable to, if not superior to, those of the most skilled human players.

AI systems that can translate between languages in real time, interpret and respond to natural speech (both spoken and written), and recognize images have also been developed (being able to recognize, identify, and sort photos or images based on the content).

However, the ability to generalize knowledge or skills is still largely a human accomplishment.

Nonetheless, there is a lot of work being done in the field of general AI right now.

It will be difficult to determine when a computer develops human-level intelligence.

Several serious and hilarious tests have been suggested to determine whether a computer has reached the level of General AI.

The Turing Test is arguably the most renowned of these examinations.

A machine and a person speak in the background, as another human listens in.

The human eavesdropper must figure out which speaker is a machine and which is a human.

The machine passes the test if it can fool the human evaluator a prescribed percentage of the time.

The Coffee Test is a more fantastical test in which a machine enters a typical household and brews coffee.

It has to find the coffee machine, look for the coffee, add water, boil the coffee, and pour it into a cup.

Another is the Flat Pack Furniture Test, which involves a machine receiving, unpacking, and assembling a piece of furniture based only on the instructions supplied.

Some scientists, as well as many science fiction writers and fans, believe that once intelligent machines reach a tipping point, they will be able to improve exponentially.

AI-based beings that far exceed human capabilities might be one conceivable result.

The Singularity, or artificial superintelligence, is the point at which AI assumes control of its own self-improvement (ASI).

If ASI is achieved, it will have unforeseeable consequences for human society.

Some pundits worry that ASI would jeopardize humanity's safety and dignity.

It's up for dispute whether the Singularity will ever happen, and how dangerous it may be.

Narrow AI applications are becoming more popular across the globe.

Machine learning (ML) is at the heart of most new applications, and most AI examples in the news are connected to this subset of technology.

Traditional or conventional algorithms are not the same as machine learning programs.

In programs that cannot learn, a computer programmer actively adds code to account for every action of an algorithm.

All of the decisions made along the process are governed by the programmer's guidelines.

This necessitates the programmer imagining and coding for every possible circumstance that an algorithm may face.

This kind of program code is bulky and often inadequate, especially if it is updated frequently to accommodate for new or unanticipated scenarios.

The utility of hard-coded algorithms approaches its limit in cases where the criteria for optimum judgments are unclear or impossible for a human programmer to foresee.

Machine learning is the process of training a computer to detect and identify patterns via examples rather than predefined rules.

This is achieved, according to Google engineer Jason Mayes, by reviewing incredibly huge quantities of training data or participating in some other kind of programmed learning step.

New patterns may be extracted by processing the test data.

The system may then classify newly unknown data based on the patterns it has already found.

Machine learning allows an algorithm to recognize patterns or rules underlying decision-making processes on its own.

Machine learning also allows a system's output to improve over time as it gains more experience (Mayes 2017).

A human programmer continues to play a vital role in this learning process, influencing results by making choices like developing the exact learning algorithm, selecting the training data, and choosing other design elements and settings.

Machine learning is powerful once it's up and running because it can adapt and enhance its ability to categorize new data without the need for direct human interaction.

In other words, the output quality increases as the user gains experience.

Artificial intelligence is a broad word that refers to the science of making computers intelligent.

AI is a computer system that can collect data and utilize it to make judgments or solve issues, according to scientists.

Another popular scientific definition of AI is "a software program paired with hardware that can receive (or sense) inputs from the world around it, evaluate and analyze those inputs, and create outputs and suggestions without the assistance of a person." When programmers claim an AI system can learn, they're referring to the program's ability to change its own processes in order to provide more accurate outputs or predictions.

AI-based systems are now being developed and used in practically every industry, from agriculture to space exploration, and in applications ranging from law enforcement to online banking.

The methods and techniques used in computer science are always evolving, extending, and improving.

Other terminology linked to machine learning, such as reinforcement learning and neural networks, are important components of cutting-edge artificial intelligence systems.

~ Jai Krishna Ponnappan

You may also want to read more about Artificial Intelligence here.

See also: 

Embodiment, AI and; Superintelligence; Turing, Alan; Turing Test.

Further Reading:

Kelnar, David. 2016. “The Fourth Industrial Revolution: A Primer on Artificial Intelligence (AI).” Medium, December 2, 2016. https://medium.com/mmc-writes/the-fourth-industrial-revolution-a-primer-on-artificial-intelligence-ai-ff5e7fffcae1.

Kurzweil, Ray. 2005. The Singularity Is Near: When Humans Transcend Biology. New York: Viking.

Mayes, Jason. 2017. Machine Learning 101. https://docs.google.com/presentation/d/1kSuQyW5DTnkVaZEjGYCkfOxvzCqGEFzWBy4e9Uedd9k/htmlpresent.

Müller, Vincent C., and Nick Bostrom. 2016. “Future Progress in Artificial Intelligence: A Survey of Expert Opinion.” In Fundamental Issues of Artificial Intelligence, edited by Vincent C. Müller, 553–71. New York: Springer.

Russell, Stuart, and Peter Norvig. 2003. Artificial Intelligence: A Modern Approach. Englewood Cliffs, NJ: Prentice Hall.

Samuel, Arthur L. 1988. “Some Studies in Machine Learning Using the Game of Checkers I.” In Computer Games I, 335–65. New York: Springer.

Artificial Intelligence - What Is The Mac Hack IV Program?

 

Mac Hack IV, a 1967 chess software built by Richard Greenblatt, gained notoriety for being the first computer chess program to engage in a chess tournament and to play adequately against humans, obtaining a USCF rating of 1,400 to 1,500.

Greenblatt's software, written in the macro assembly language MIDAS, operated on a DEC PDP-6 computer with a clock speed of 200 kilohertz.

While a graduate student at MIT's Artificial Intelligence Laboratory, he built the software as part of Project MAC.

"Chess is the drosophila [fruit fly] of artificial intelligence," according to Russian mathematician Alexander Kronrod, the field's chosen experimental organ ism (Quoted in McCarthy 1990, 227).

Creating a champion chess software has been a cherished goal in artificial intelligence since 1950, when Claude Shan ley first described chess play as a task for computer programmers.

Chess and games in general involve difficult but well-defined issues with well-defined rules and objectives.

Chess has long been seen as a prime illustration of human-like intelligence.

Chess is a well-defined example of human decision-making in which movements must be chosen with a specific purpose in mind, with limited knowledge and uncertainty about the result.

The processing capability of computers in the mid-1960s severely restricted the depth to which a chess move and its alternative answers could be studied since the number of different configurations rises exponentially with each consecutive reply.

The greatest human players have been proven to examine a small number of moves in greater detail rather than a large number of moves in lower depth.

Greenblatt aimed to recreate the methods used by good players to locate significant game tree branches.

He created Mac Hack to reduce the number of nodes analyzed while choosing moves by using a minimax search of the game tree along with alpha-beta pruning and heuristic components.

In this regard, Mac Hack's style of play was more human-like than that of more current chess computers (such as Deep Thought and Deep Blue), which use the sheer force of high processing rates to study tens of millions of branches of the game tree before making moves.

In a contest hosted by MIT mathematician Seymour Papert in 1967, Mac Hack defeated MIT philosopher Hubert Dreyfus and gained substantial renown among artificial intelligence researchers.

The RAND Corporation published a mimeographed version of Dreyfus's paper, Alchemy and Artificial Intelligence, in 1965, which criticized artificial intelligence researchers' claims and aspirations.

Dreyfus claimed that no computer could ever acquire intelligence since human reason and intelligence are not totally rule-bound, and hence a computer's data processing could not imitate or represent human cognition.

In a part of the paper titled "Signs of Stagnation," Dreyfus highlighted attempts to construct chess-playing computers, among his many critiques of AI.

Mac Hack's victory against Dreyfus was first seen as vindication by the AI community.

~ Jai Krishna Ponnappan

You may also want to read more about Artificial Intelligence here.

See also: 

Alchemy and Artificial Intelligence; Deep Blue.

Further Reading:

Crevier, Daniel. 1993. AI: The Tumultuous History of the Search for Artificial Intelligence. New York: Basic Books.

Greenblatt, Richard D., Donald E. Eastlake III, and Stephen D. Crocker. 1967. “The Greenblatt Chess Program.” In AFIPS ’67: Proceedings of the November 14–16, 1967, Fall Joint Computer Conference, 801–10. Washington, DC: Thomson Book Company.

Marsland, T. Anthony. 1990. “A Short History of Computer Chess.” In Computers, Chess, and Cognition, edited by T. Anthony Marsland and Jonathan Schaeffer, 3–7. New York: Springer-Verlag.

McCarthy, John. 1990. “Chess as the Drosophila of AI.” In Computers, Chess, and Cognition, edited by T. Anthony Marsland and Jonathan Schaeffer, 227–37. New York: Springer-Verlag.

McCorduck, Pamela. 1979. Machines Who Think: A Personal Inquiry into the History and Prospects of Artificial Intelligence. San Francisco: W. H. Freeman.