Talk:Artificial intelligence/Where did it go? 2021

In the summer and fall of 2021, I copy-edited the entire article for redundancy, WP:RELEVANCE, WP:UNDUE weight, organization and citation format. Most of the material was moved to sub-articles, such as applications of AI, artificial general intelligence, history of AI and so on. Some material (marked "Not Done" below) didn't seem to fit in anywhere, or was difficult to save for one reason or another. ---- CharlesGillingham (talk) 23:16, 6 October 2021 (UTC)

From History

 Done These have been moved to Applications of AI. All but three of these have a one sentence mention in Artificial intelligence § Appliations --- CharlesGillingham (talk) 16:31, 29 September 2021 (UTC)

• Deep Blue became the first computer chess-playing system to beat a reigning world chess champion, Garry Kasparov, on 11 May 1997.
• In 2011, in a Jeopardy! quiz show exhibition match, IBM's question answering system, Watson, defeated the two greatest Jeopardy! champions, Brad Rutter and Ken Jennings, by a significant margin.
• In March 2016, AlphaGo won 4 out of 5 games of Go in a match with Go champion Lee Sedol, becoming the first computer Go-playing system to beat a professional Go player without handicaps. In the 2017 Future of Go Summit, AlphaGo won a three-game match with Ke Jie, who at the time continuously held the world No. 1 ranking for two years. Deep Blue's Murray Campbell called AlphaGo's victory "the end of an era... board games are more or less done and it's time to move on." This marked the completion of a significant milestone in the development of Artificial Intelligence as Go is a relatively complex game, more so than Chess. AlphaGo was later improved, generalized to other games like chess, with AlphaZero; and MuZero to play many different video games, that were previously handled separately, in addition to board games.
• Other programs handle imperfect-information games; such as for poker at a superhuman level, Pluribus (poker bot) and Cepheus (poker bot). See: General game playing.
• Intelligent personal assistants, such as Siri and Alexa, are able to understand many natural language requests.
• Microsoft developed a Skype system that can automatically translate from one language to another.
• DeepMind's AlphaFold 2 (2020) demonstrated the ability to determine, in hours rather than months, the 3D structure of a protein. Facial recognition advanced to where, under some circumstances, some systems claim to have a 99% accuracy rate.
• Facebook developed a system that can describe images to blind people.
• The Kinect, which provides a 3D body–motion interface for the Xbox 360 and the Xbox One, uses algorithms that emerged from lengthy AI research

 Done Moved to Artificial intelligence § Applications

By 2020, Natural Language Processing systems such as the enormous GPT-3 (then by far the largest artificial neural network) were matching human performance on pre-existing benchmarks, albeit without the system attaining commonsense understanding of the contents of the benchmarks.

 Not done China's AI program is not (yet) the most important trend of the decade. Perhaps the paragraph on the 2020s will use this. ---- CharlesGillingham (talk) 18:44, 12 September 2021 (UTC)

Around 2016, China greatly accelerated its government funding; given its large supply of data and its rapidly increasing research output, some observers believe it may be on track to becoming an "AI superpower".

From Basics

The article had a section called Basics which was an article-within-the-article. This is very well written, well sourced and accurate, but it is completely redundant. We still need to look at the best bits and see if they aren't better than what we already have on those topics and replace what we have if that's a good idea.

 Done Moved to intelligent agent

Computer science defines AI research as the study of "intelligent agents": any device that perceives its environment and takes actions that maximize its chance of successfully achieving its goals. A more elaborate definition characterizes AI as "a system's ability to correctly interpret external data, to learn from such data, and to use those learnings to achieve specific goals and tasks through flexible adaptation."

 Done Moved to intelligent agent

A typical AI analyzes its environment and takes actions that maximize its chance of success. An AI's intended utility function (or goal) can be simple ("1 if the AI wins a game of Go, 0 otherwise") or complex ("Perform actions mathematically similar to ones that succeeded in the past"). Goals can be explicitly defined or induced. If the AI is programmed for "reinforcement learning", goals can be implicitly induced by rewarding some types of behavior or punishing others. Alternatively, an evolutionary system can induce goals by using a "fitness function" to mutate and preferentially replicate high-scoring AI systems, similar to how animals evolved to innately desire certain goals such as finding food. Some AI systems, such as nearest-neighbor, instead of reason by analogy, these systems are not generally given goals, except to the degree that goals are implicit in their training data. Such systems can still be benchmarked if the non-goal system is framed as a system whose "goal" is to accomplish its narrow classification task.

 Not done Where? Algorithm?

AI often revolves around the use of algorithms. An algorithm is a set of unambiguous instructions that a mechanical computer can execute. A complex algorithm is often built on top of other, simpler, algorithms. A simple example of an algorithm is the following (optimal for first player) recipe for play at tic-tac-toe:

1. If someone has a "threat" (that is, two in a row), take the remaining square. Otherwise,
2. if a move "forks" to create two threats at once, play that move. Otherwise,
3. take the center square if it is free. Otherwise,
4. if your opponent has played in a corner, take the opposite corner. Otherwise,
5. take an empty corner if one exists. Otherwise,
6. take any empty square.

TODO Heuristic learning?

Many AI algorithms are capable of learning from data; they can enhance themselves by learning new heuristics (strategies, or "rules of thumb", that have worked well in the past), or can themselves write other algorithms.

 Not done Dubious.

Some of the "learners" described below, including Bayesian networks, decision trees, and nearest-neighbor, could theoretically, (given infinite data, time, and memory) learn to approximate any function,including which combination of mathematical functions would best describe the world. These learners could therefore derive all possible knowledge, by considering every possible hypothesis and matching them against the data.

TODO Move to Intractability (just the example)

In practice, it is seldom possible to consider every possibility, because of the phenomenon of "combinatorial explosion", where the time needed to solve a problem grows exponentially. Much of AI research involves figuring out how to identify and avoid considering a broad range of possibilities unlikely to be beneficial. For example, when viewing a map and looking for the shortest driving route from Denver to New York in the East, one can in most cases skip looking at any path through San Francisco or other areas far to the West; thus, an AI wielding a pathfinding algorithm like A* can avoid the combinatorial explosion that would ensue if every possible route had to be ponderously considered.

 Not done This is really good, especially the examples, but I'm not sure where to work it into the article or anywhere else in Wikipedia. The Tools section basically covers these same points in the same order. Could it work there?

The earliest (and easiest to understand) approach to AI was symbolism (such as formal logic): "If an otherwise healthy adult has a fever, then they may have influenza".

A second, more general, approach is Bayesian inference: "If the current patient has a fever, adjust the probability they have influenza in such-and-such way".

The third major approach, extremely popular in routine business AI applications, are analogizers such as SVM and nearest-neighbor: "After examining the records of known past patients whose temperature, symptoms, age, and other factors mostly match the current patient, X% of those patients turned out to have influenza".

A fourth approach is harder to intuitively understand, but is inspired by how the brain's machinery works: the artificial neural network approach uses artificial "neurons" that can learn by comparing itself to the desired output and altering the strengths of the connections between its internal neurons to "reinforce" connections that seemed to be useful. These four main approaches can overlap with each other and with evolutionary systems; for example, neural nets can learn to make inferences, to generalize, and to make analogies. Some systems implicitly or explicitly use multiple of these approaches, alongside many other AI and non-AI algorithms; the best approach is often different depending on the problem.

 Done Moved to Machine learning

Learning algorithms work on the basis that strategies, algorithms, and inferences that worked well in the past are likely to continue working well in the future. These inferences can be obvious, such as "since the sun rose every morning for the last 10,000 days, it will probably rise tomorrow morning as well". They can be nuanced, such as "X% of families have geographically separate species with color variants, so there is a Y% chance that undiscovered black swans exist".

 Done I'm not confident about where this fits into machine learning, so I can't put it anywhere myself. Sending it to Talk:Machine learning.

Learners can also work on the basis of "Occam's razor": The simplest theory that explains the data is the likeliest. Therefore, according to Occam's razor principle, a learner must be designed such that it prefers simpler theories to complex theories, except in cases where the complex theory is proven substantially better.

 Done Moved to Machine learning § Limitations

The blue line could be an example of overfitting a linear function due to random noise.

Settling on a bad, overly complex theory gerrymandered to fit all the past training data is known as overfitting. Many systems attempt to reduce overfitting by rewarding a theory in accordance with how well it fits the data, but penalizing the theory in accordance with how complex the theory is. Besides classic overfitting, learners can also disappoint by "learning the wrong lesson". A toy example is that an image classifier trained only on pictures of brown horses and black cats might conclude that all brown patches are likely to be horses. A real-world example is that, unlike humans, current image classifiers often don't primarily make judgments from the spatial relationship between components of the picture, and they learn relationships between pixels that humans are oblivious to, but that still correlate with images of certain types of real objects. Modifying these patterns on a legitimate image can result in "adversarial" images that the system misclassifies.

 Done Point is already made in Artificial intelligence § knowledge. This text appears in Commonsense reasoning.

A self-driving car system may use a neural network to determine which parts of the picture seem to match previous training images of pedestrians, and then model those areas as slow-moving but somewhat unpredictable rectangular prisms that must be avoided.

AI lacks several features of human "commonsense reasoning"; most notably, humans have powerful mechanisms for reasoning about "naïve physics" such as space, time, and physical interactions. This enables even young children to easily make inferences like "If I roll this pen off a table, it will fall on the floor". Humans also have a powerful mechanism of "folk psychology" that helps them to interpret natural-language sentences such as "The city councilmen refused the demonstrators a permit because they advocated violence" (A generic AI has difficulty discerning whether the ones alleged to be advocating violence are the councilmen or the demonstrators).

This lack of "common knowledge" means that AI often makes different mistakes than humans make, in ways that can seem incomprehensible. For example, existing self-driving cars cannot reason about the location nor the intentions of pedestrians in the exact way that humans do, and instead must use non-human modes of reasoning to avoid accidents.

From Goals

From Goals/Lede

 Not done Is this a re-invention/re-framing of symbolic vs. sub-symbolic? Perhaps it could go in symbolic AI; although I would really like to see this in a WP:SECONDARY source. ---- CharlesGillingham (talk) 04:18, 6 October 2021 (UTC)

The cognitive capabilities of current architectures are very limited, using only a simplified version of what intelligence is really capable of. For instance, the human mind has come up with ways to reason beyond measure and logical explanations to different occurrences in life. What would have been otherwise straightforward, an equivalently difficult problem may be challenging to solve computationally as opposed to using the human mind. This gives rise to two classes of models: structuralist and functionalist. The structural models aim to loosely mimic the basic intelligence operations of the mind such as reasoning and logic. The functional model refers to the correlating data to its computed counterpart.

From Social Intelligence

 Done The first source is actually about technological employment (the paradox is relevant because computers are bad at perceptual and motor tasks). Moved the source to artificial intelligence § Technological unemployment. The second source is about giving AI programs a "theory of (other) minds", which is a form of social intelligence. Added the citation to Artificial intelligence § Social intelligence ---- CharlesGillingham (talk) 22:09, 6 October 2021 (UTC)

Moravec's paradox can be extended to many forms of social intelligence.

 Not done Too vague to be useful anywhere.

Distributed multi-agent coordination of autonomous vehicles remains a difficult problem.

From General Intelligence

 Done Cyc is covered in the article History of AI as well as Artificial general intelligence, FGCP is covered in a footnote in Artificial intelligence § History (UTC)

Historically, projects such as the Cyc knowledge base (1984–) and the massive Japanese Fifth Generation Computer Systems initiative (1982–1992) attempted to cover the breadth of human cognition. These early projects failed to escape the limitations of non-quantitative symbolic logic models and, in retrospect, greatly underestimated the difficulty of cross-domain AI.

 Done This has been moved to Applications of AI

One high-profile example is that DeepMind in the 2010s developed a "generalized artificial intelligence" that could learn many diverse Atari games on its own, and later developed a variant of the system which succeeds at sequential learning.

 Done This is added to artificial intelligence § Learning

Besides transfer learning,

 Not done This is unclear. However, the source is perfect and the point is good. Needs layman's language (first half) and encyclopedic tone (second half)

hypothetical AGI breakthroughs could include the development of reflective architectures that can engage in decision-theoretic metareasoning, and figuring out how to "slurp up" a comprehensive knowledge base from the entire unstructured Web.

 Done Moved to artificial general intelligence

Many of the problems in this article may also require general intelligence, if machines are to solve the problems as well as people do. For example, even specific straightforward tasks, like machine translation, require that a machine read and write in both languages (NLP), follow the author's argument (reason), know what is being talked about (knowledge), and faithfully reproduce the author's original intent (social intelligence). A problem like machine translation is considered "AI-complete", because all of these problems need to be solved simultaneously in order to reach human-level machine performance.

From "Approaches"

Before 2021, the article had a section called "Approaches". This has been divided between History, Philosophy and the sub-articles. ---- CharlesGillingham (talk) 18:21, 12 September 2021 (UTC)

From Symbolic AI

 Done I moved this section into Symbolic AI. The history Symbolic AI of is described in two paragraphs of Artificial Intelligence § History, and the weaknesses and strengths of the approach are describe in the section Artificial intelligence § Symbolic AI and its limits

During the 1960s, symbolic approaches achieved great success at simulating intelligent behavior in small demonstration programs. AI research was centered in three institutions in the 1960s: Carnegie Mellon University, Stanford, MIT and (later) University of Edinburgh. Each one developed its own style of research. Earlier approaches based on cybernetics or artificial neural networks were abandoned or pushed into the background.

Cognitive simulation

Economist Herbert Simon and Allen Newell studied human problem-solving skills and attempted to formalize them, and their work laid the foundations of the field of artificial intelligence, as well as cognitive science, operations research and management science. Their research team used the results of psychological experiments to develop programs that simulated the techniques that people used to solve problems. This tradition, centered at Carnegie Mellon University would eventually culminate in the development of the Soar architecture in the middle 1980s.

Logic-based

Main article: logic programming

Unlike Simon and Newell, John McCarthy felt that machines did not need to simulate human thought, but should instead try to find the essence of abstract reasoning and problem-solving, regardless of whether people used the same algorithms. His laboratory at Stanford (SAIL) focused on using formal logic to solve a wide variety of problems, including knowledge representation, planning and learning. Logic was also the focus of the work at the University of Edinburgh and elsewhere in Europe which led to the development of the programming language Prolog and the science of logic programming.

Anti-logic or "scruffy"

Main article: neats vs. scruffies

Researchers at MIT (such as Marvin Minsky and Seymour Papert) found that solving difficult problems in vision and natural language processing required ad hoc solutions—they argued that no simple and general principle (like logic) would capture all the aspects of intelligent behavior. Roger Schank described their "anti-logic" approaches as "scruffy" (as opposed to the "neat" paradigms at CMU and Stanford). Commonsense knowledge bases (such as Doug Lenat's Cyc) are an example of "scruffy" AI, since they must be built by hand, one complicated concept at a time.

Knowledge based

When computers with large memories became available around 1970, researchers from all three traditions began to build knowledge into AI applications. The knowledge revolution was driven by the realization that enormous amounts of knowledge would be required by many simple AI applications.

From Embodied Intelligence

 Done The coverage is sufficient, and of course this definition is in the article developmental robotics. --- CharlesGillingham (talk) 03:02, 1 October 2021 (UTC)

Within developmental robotics, developmental learning approaches are elaborated upon to allow robots to accumulate repertoires of novel skills through autonomous self-exploration, social interaction with human teachers, and the use of guidance mechanisms (active learning, maturation, motor synergies, etc.).

From Integrating the Approaches

TODO Artificial intelligence § General intelligence mentions cognitive architectures and multi-agent systems as approaches to AGI, and the others here are mentioned in a footnote. Technically, I can't call this "Done" because our article doesn't acknowledge that these are also used as tools for particular applications. Still might need to have a (very short) section on this stuff in Tools.

• Agent architectures and cognitive architectures: Researchers have designed systems to build intelligent systems out of interacting intelligent agents in a multi-agent system.
• A hierarchical control system provides a bridge between sub-symbolic AI at its lowest, reactive levels and traditional symbolic AI at its highest levels, where relaxed time constraints permit planning and world modeling.
• Some cognitive architectures are custom-built to solve a narrow problem; others, such as Soar, are designed to mimic human cognition and to provide insight into general intelligence.
• Modern extensions of Soar are hybrid intelligent systems that include both symbolic and sub-symbolic components.

From Tools

From Logic

 Done These points have been moved into Fuzzy logic#Applications.

Fuzzy logic is successfully used in control systems to allow experts to contribute vague rules such as "if you are close to the destination station and moving fast, increase the train's brake pressure"; these vague rules can then be numerically refined within the system. Fuzzy logic fails to scale well in knowledge bases; many AI researchers question the validity of chaining fuzzy-logic inferences.

From neural networks

 Not done I think that the author of this was trying explain that information is distributed throughout the network, rather than being stored in a specific location (as it would be with symbolic AI). however using the word "concepts" (which has a specific meaning in cognitive science) is a misleading way to describe this -- it actually confuses the issue. This is also unsourced. Perhaps someone else can figure out what the original author meant and say it better.

The neural network forms "concepts" that are distributed among a subnetwork of shared neurons that tend to fire together; a concept meaning "leg" might be coupled with a subnetwork meaning "foot" that includes the sound for "foot".

 Done Moved to Artificial neural network § History

Neural networks' early successes included predicting the stock market and (in 1995) a mostly self-driving car.

 Done This point is made Artificial intelligence § History

In the 2010s, advances in neural networks using deep learning thrust AI into widespread public consciousness and contributed to an enormous upshift in corporate AI spending;

 Not done Artificial intelligence § History already reports two excellent metrics of the uptick in AI interest 2015-2020 (total publications, corporate spending). This is not a particularly notable metric, and we can't really use it when we have better ones.

AI-related M&A in 2017 was over 25 times as large as in 2015.

 Done Frank Rosenblatt is discussed in the History of AI, and Pitts & McCullough is mentioned there and in Artificial intelligence § History.

The study of non-learning neural networks began in the decade before the field of AI research was founded, in the work of Walter Pitts and Warren McCullouch. Frank Rosenblatt invented the perceptron, a learning network with a single layer, similar to the old concept of linear regression.

 Not done without sources, Wikipedia can't really make any assertion about their importance.

Early pioneers also include Alexey Grigorevich Ivakhnenko, Teuvo Kohonen, Stephen Grossberg, Kunihiko Fukushima, Christoph von der Malsburg, David Willshaw, Shun-Ichi Amari, Bernard Widrow, John Hopfield, Eduardo R. Caianiello, and others

 Done Linnaimaa is credited in a footnote.

which has been around since 1970 as the reverse mode of automatic differentiation published by Seppo Linnainmaa, and was introduced to neural networks by Paul Werbos.

 Not done WP:UNDUE weight on this approach. Can't really move this to an AI sub-article either, because it's not really in use -- biologically based AI, maybe?

Hierarchical temporal memory is an approach that models some of the structural and algorithmic properties of the neocortex.

 Done Similarly, this is probably WP:UNDUE weight on this approach. Moved to Artificial neural network

However, some research groups, such as Uber, argue that simple neuroevolution to mutate new neural network topologies and weights may be competitive with sophisticated gradient descent approaches. One advantage of neuroevolution is that it may be less prone to get caught in "dead ends".

From Feedforward Networks

 Done All this precedence is covered in Deep learning

According to one overview,the expression "Deep Learning" was introduced to the machine learning community by Rina Dechter in 1986 and gained traction after Igor Aizenberg and colleagues introduced it to artificial neural networks in 2000. The first functional deep Learning networks were published by Alexey Grigorevich Ivakhnenko and V. G. Lapa in 1965. These networks are trained one layer at a time. Ivakhnenko's 1971 paper describes the learning of a deep feedforward multilayer perceptron with eight layers, already much deeper than many later networks.

 Done Too much undefined WP:JARGON. Significance isn't clear. Covered in Deep learning.

In 2006, a publication by Geoffrey Hinton and Ruslan Salakhutdinov introduced another way of pre-training many-layered feedforward neural networks (FNNs) one layer at a time, treating each layer in turn as an unsupervised restricted Boltzmann machine, then using supervised backpropagation for fine-tuning. Similar to shallow artificial neural networks, deep neural networks can model complex non-linear relationships.

 Done More precedence. Covered in Deep learning.

(CNNs), whose origins can be traced back to the Neocognitron introduced by Kunihiko Fukushima in 1980. In 1989, Yann LeCun and colleagues applied backpropagation to such an architecture.

 Done Covered in Deep learning

In the early 2000s, in an industrial application, CNNs already processed an estimated 10% to 20% of all the checks written in the US. Since 2011, fast implementations of CNNs on GPUs have won many visual pattern recognition competitions.

 Done The article has enough detail about AlphaGo. Moved to AlphaGo.

CNNs with 12 convolutional layers were used with reinforcement learning by Deepmind's "AlphaGo Lee", the program that beat a top Go champion in 2016.

From Deep recurrent neural networks

Main article: Recurrent neural networks

 Not done This is unsourced (but unlikely to be challenged). Still, don't think we need it, since there are more applications today.

recurrent neural networks have been used for sequence learning.

 Done Just the source.

(RNNs)

 Done Moved to Recurrent neural networks, where this fact did not appear (and thus probably not notable enough for this article).

recurrent neural network are theoretically Turing complete and can run arbitrary programs to process arbitrary sequences of inputs.

 Done Kept, Edited for brevity.

The depth of an RNN is unlimited and depends on the length of its input sequence; thus, an RNN is an example of deep learning.

 Done Schmidhuber's work 1991-92 is described in Recurrent neural network.

In 1992, it was shown that unsupervised pre-training of a stack of recurrent neural networks can speed up subsequent supervised learning of deep sequential problems.

 Done LSTM is mentioned, with this source.

Numerous researchers now use variants of a deep learning recurrent NN called the long short-term memory (LSTM) network published by Hochreiter & Schmidhuber in 1997.

 Done Undefined WP:JARGON. This is covered in Recurrent Neural Network.

LSTM is often trained by Connectionist Temporal Classification (CTC).

 Done Applications of LSTM. These projects are described in Recurrent neural network § LSTM, with the same sources.

At Google, Microsoft and Baidu this approach has revolutionized speech recognition. For example, in 2015, Google's speech recognition experienced a dramatic performance jump of 49% through CTC-trained LSTM. Google also used LSTM to improve machine translation, language modeling, and multilingual language processing. LSTM combined with CNNs also improved automatic image captioning and a plethora of other applications.

From Applications

 Done Moved to Applications of AI

With social media sites overtaking TV as a source for news for young people and news organizations increasingly reliant on social media platforms for generating distribution,

From Evaluating progress

 Done Moved into Applications of AI.

AI, like electricity or the steam engine, is a general purpose technology. There is no consensus on how to characterize which tasks AI tends to excel at.

 Done Moved into Applications of AI.

While projects such as AlphaZero have succeeded in generating their own knowledge from scratch, many other machine learning projects require large training datasets.

 Done Moved into Moravec's paradox. ---- CharlesGillingham (talk) 16:41, 12 October 2021 (UTC)

Researcher Andrew Ng has suggested, as a "highly imperfect rule of thumb", that "almost anything a typical human can do with less than one second of mental thought, we can probably now or in the near future automate using AI."

 Done Moravec's paradox is covered Artificial intelligence § Symbolic AI and its limits

Moravec's paradox suggests that AI lags humans at many tasks that the human brain has specifically evolved to perform well.

 Done Games & AlphaGo are covered in Artificial intelligence § Applications

Games provide a well-publicized benchmark for assessing rates of progress. AlphaGo around 2016 brought the era of classical board-game benchmarks to a close.

 Done This appears in Progress in artificial intelligence.

Games of imperfect knowledge provide new challenges to AI in game theory.

 Done This is moved to Applications of AI.

E-sports such as StarCraft continue to provide additional public benchmarks.

 Done This has been added to Progress in artificial intelligence

Many competitions and prizes, such as the Imagenet Challenge, promote research in artificial intelligence. The most common areas of competition include general machine intelligence, conversational behavior, data-mining, robotic cars, and robot soccer as well as conventional games.

 Done This appears in Progress in artificial intelligence

The "imitation game" (an interpretation of the 1950 Turing test that assesses whether a computer can imitate a human) is nowadays considered too exploitable to be a meaningful benchmark. A derivative of the Turing test is the Completely Automated Public Turing test to tell Computers and Humans Apart (CAPTCHA). As the name implies, this helps to determine that a user is an actual person and not a computer posing as a human. Unlike the standard Turing test, CAPTCHA is administered by a machine and targeted to a human as opposed to being administered by a human and targeted to a machine. A computer asks a user to complete a simple test then generates a grade for that test. Computers are unable to solve the problem, so correct solutions are deemed to be the result of a person taking the test. A common type of CAPTCHA is the test that requires the typing of distorted letters, numbers or symbols that appear in an image undecipherable by a computer.

 Done This appears in Progress in artificial intelligence

Proposed "universal intelligence" tests aim to compare how well machines, humans, and even non-human animals perform on problem sets that are generic as possible. At an extreme, the test suite can contain every possible problem, weighted by Kolmogorov complexity; unfortunately, these problem sets tend to be dominated by impoverished pattern-matching exercises where a tuned AI can easily exceed human performance levels.

 Done Moved to Hardware for artificial intelligence

Since the 2010s, advances in both machine learning algorithms and computer hardware have led to more efficient methods for training deep neural networks that contain many layers of non-linear hidden units and a very large output layer. By 2019, graphic processing units (GPUs), often with AI-specific enhancements, had displaced CPUs as the dominant method of training large-scale commercial cloud AI. OpenAI estimated the hardware compute used in the largest deep learning projects from AlexNet (2012) to AlphaZero (2017), and found a 300,000-fold increase in the amount of compute required, with a doubling-time trendline of 3.4 months.

From Philosophy

 Done This is covered in Philosophy of AI

In the proposal for the Dartmouth Workshop of 1956, John McCarthy wrote "Every aspect of learning or any other feature of intelligence can be so precisely described that a machine can be made to simulate it."

 Done This is covered in Philosophy of AI

Kurt Gödel, John Lucas (in 1961) and Roger Penrose (in a more detailed argument from 1989 onwards) made highly technical arguments that human mathematicians can consistently see the truth of their own "Gödel statements" and therefore have computational abilities beyond that of mechanical Turing machines. However, some people do not agree with the "Gödelian arguments".

 Done The AI effect has been covered in the Lede and in Applications.

The AI effect claims that machines are already intelligent, but observers have failed to recognize it. For example, when Deep Blue beat Garry Kasparov in chess, the machine could be described as exhibiting intelligence. However, onlookers commonly discount the behavior of an artificial intelligence program by arguing that it is not "real" intelligence, with "real" intelligence being in effect defined as whatever behavior machines cannot do.

 Done This is covered in philosophy of AI

The artificial brain argument asserts that the brain can be simulated by machines and, because brains exhibit intelligence, these simulated brains must also exhibit intelligence − ergo, machines can be intelligent. Hans Moravec, Ray Kurzweil and others have argued that it is technologically feasible to copy the brain directly into hardware and software, and that such a simulation will be essentially identical to the original.

From Future of AI

From Singularity

 Done Kurzweil's prediction is covered in artificial general intelligence

Ray Kurzweil has used Moore's law (which describes the relentless exponential improvement in digital technology) to calculate that desktop computers will have the same processing power as human brains by the year 2029 and predicts that the singularity will occur in 2045.

From Robot Rights

 Done Plug & Play is mentioned in the footnote

The subject is profoundly discussed in the 2010 documentary film Plug & Pray

 Not done Can't really move this into artificial intelligence in fiction because that article is tightly structured and there's no place for this topic at the moment.

and many sci fi media such as Star Trek Next Generation, with the character of Commander Data, who fought being disassembled for research, and wanted to "become human", and the robotic holograms in Voyager.

From Risks

 Not done This is devoid of actual content about AI, and too vague to be useful in existential risk of artificial intelligence

The potential negative effects of AI and automation were a major issue for Andrew Yang's 2020 presidential campaign in the United States.

 Not done Redundant. The points that Beridze is making are vague and are covered in more detail elsewhere in the article. Added this citation to a paragraph about the same concerns citing Musk, Gates and Hawkins. Also a bit vague to be useful in Existential risk of AI

Irakli Beridze, Head of the Centre for Artificial Intelligence and Robotics at UNICRI, United Nations, has expressed that "I think the dangerous applications for AI, from my point of view, would be criminals or large terrorist organizations using it to disrupt large processes or simply do pure harm. [Terrorists could cause harm] via digital warfare, or it could be a combination of robotics, drones, with AI and other things as well that could be really dangerous. And, of course, other risks come from things like job losses. If we have massive numbers of people losing jobs and don't find a solution, it will be extremely dangerous. Things like lethal autonomous weapons systems should be properly governed—otherwise there's massive potential of misuse."

From technological unemployment

 Done Redundant: Each contribution seemed to want to introduce the topic again.

• The long-term economic effects of AI are uncertain.
• about whether the increasing use of robots and AI will cause a substantial increase in long-term unemployment

 Done Redundant: This point was made twice, and I chose the one based on Ford. Keeping the reference.

The relationship between automation and employment is complicated. While automation eliminates old jobs, it also creates new jobs through micro-economic and macro-economic effects.

 Not done These were off-topic

• A 2017 study by PricewaterhouseCoopers sees the People's Republic of China gaining economically the most out of AI with 26.1% of GDP until 2030.
• A February 2020 European Union white paper on artificial intelligence advocated for artificial intelligence for economic benefits, including "improving healthcare (e.g. making diagnosis more precise, enabling better prevention of diseases), increasing the efficiency of farming, contributing to climate change mitigation and adaptation, [and] improving the efficiency of production systems through predictive maintenance", while acknowledging potential risks.

 Done Moved to technological unemployment

Author Martin Ford and others go further and argue that many jobs are routine, repetitive and (to an AI) predictable; Ford warns that these jobs may be automated in the next couple of decades, and that many of the new jobs may not be "accessible to people with average capability", even with retraining.

From Existential Risk

 Done Kept a sentence of this. This point is also made in Existential risk of AI (in three places).

Physicist Stephen Hawking, Microsoft founder Bill Gates, history professor Yuval Noah Harari, and SpaceX founder Elon Musk have expressed concerns about the possibility that AI could evolve to the point that humans could not control it, with Hawking theorizing that this could "spell the end of the human race".

The development of full artificial intelligence could spell the end of the human race. Once humans develop artificial intelligence, it will take off on its own and redesign itself at an ever-increasing rate. Humans, who are limited by slow biological evolution, couldn't compete and would be superseded.

— Stephen Hawking

 Done Kept one sentence of each of this, the whole paragraph is moved to Existential risk of AI

In his book Superintelligence, philosopher Nick Bostrom provides an argument that artificial intelligence will pose a threat to humankind. He argues that sufficiently intelligent AI, if it chooses actions based on achieving some goal, will exhibit convergent behavior such as acquiring resources or protecting itself from being shut down. If this AI's goals do not fully reflect humanity's—one example is an AI told to compute as many digits of pi as possible—it might harm humanity in order to acquire more resources or prevent itself from being shut down, ultimately to better achieve its goal.

 Done ame deal. Summary in AI, all the text moved to Existential risk of AI

Bostrom also emphasizes the difficulty of fully conveying humanity's values to an advanced AI. He uses the hypothetical example of giving an AI the goal to make humans smile to illustrate a misguided attempt. If the AI in that scenario were to become superintelligent, Bostrom argues, it may resort to methods that most humans would find horrifying, such as inserting "electrodes into the facial muscles of humans to cause constant, beaming grins" because that would be an efficient way to achieve its goal of making humans smile.

 Done This is covered in Friendly AI

In his book Human Compatible AI researcher Stuart J. Russell echoes some of Bostrom's concerns while also proposing an approach to developing provably beneficial machines focused on uncertainty and deference to humans possibly involving inverse reinforcement learning.

 Done Kept one sentence or so from this, the entire paragraph moved to Existential risk of AI

Concern over risk from artificial intelligence has led to some high-profile donations and investments. A group of prominent tech titans including Peter Thiel, Amazon Web Services and Musk have committed $1 billion to OpenAI, a nonprofit company aimed at championing responsible AI development. In January 2015, Elon Musk donated $10 million to the Future of Life Institute to fund research on understanding AI decision making. The goal of the institute is to "grow wisdom with which we manage" the growing power of technology. Musk also funds companies developing artificial intelligence such as DeepMind and Vicarious to "just keep an eye on what's going on with artificial intelligence. I think there is potentially a dangerous outcome there."

 Done This is moved to Existential risk of AI

The opinion of experts within the field of artificial intelligence is mixed, with sizable fractions both concerned and unconcerned by risk from eventual superhumanly-capable AI.

 Done This is moved to Technological unemployment

Oracle CEO Mark Hurd has stated that AI "will actually create more jobs, not less jobs" as humans will be needed to manage AI systems.

 Done This is in Existential risk of AI

Facebook CEO Mark Zuckerberg believes AI will "unlock a huge amount of positive things," such as curing disease and increasing the safety of autonomous cars.

 Done This is in Existential risk of AI

For the danger of uncontrolled advanced AI to be realized, the hypothetical AI would have to overpower or out-think all of humanity, which a minority of experts argue is a possibility far enough in the future to not be worth researching.

 Done This is in Existential risk of AI.

Other counterarguments revolve around humans being either intrinsically or convergently valuable from the perspective of an artificial intelligence.

From Ethical machines

 Done Everything here is either in ethics of AI or history of AI

Joseph Weizenbaum in Computer Power and Human Reason wrote that AI applications cannot, by definition, successfully simulate genuine human empathy and that the use of AI technology in fields such as customer service or psychotherapy was deeply misguided. Weizenbaum was also bothered that AI researchers (and some philosophers) were willing to view the human mind as nothing more than a computer program (a position now known as computationalism). To Weizenbaum, these points suggest that AI research devalues human life.

From Malevolent AI

 Done A shortened version of this paragraph was moved up into the "weaponized A" section.

Lethal autonomous weapons are of concern. By 2015, over fifty countries were reported to be researching battlefield robots, including the United States, China, Russia, and the United Kingdom. Many people concerned about risk from superintelligent AI also want to limit the use of artificial soldiers and drones.

 Done Added this citation and footnote with the quote to the "existential risk" section, because this is a response to the risk. Also added the full quote to Existential risk of AI

Leading AI researcher Rodney Brooks writes, "I think it is a mistake to be worrying about us developing malevolent AI anytime in the next few hundred years. I think the worry stems from a fundamental error in not distinguishing the difference between the very real recent advances in a particular aspect of AI and the enormity and complexity of building sentient volitional intelligence."

 DoneMoved (a cut-down version of) this into "existential risk" because it is an argument that there is a risk. The remainder of this was moved into Existential risk of AI § Orthogonality thesis

Political scientist Charles T. Rubin believes that AI can be neither designed nor guaranteed to be benevolent. He argues that "any sufficiently advanced benevolence may be indistinguishable from malevolence." Humans should not assume machines or robots would treat us favorably because there is no a priori reason to believe that they would be sympathetic to our system of morality, which has evolved along with our particular biology (which AIs would not share). Hyper-intelligent software may not necessarily decide to support the continued existence of humanity and would be extremely difficult to stop. This topic has also recently begun to be discussed in academic publications as a real source of risks to civilization, humans, and planet Earth.

From Regulation

 Done All of these paragraphs (or equivalent) and their sources now appear in regulation of AI

Regulation of AI through mechanisms such as review boards can also be seen as social means to approach the AI control problem.

The Global Partnership on Artificial Intelligence was launched in June 2020, stating a need for AI to be developed in accordance with human rights and democratic values, to ensure public confidence and trust in the technology, as outlined in the OECD Principles on Artificial Intelligence (2019). The founding members of the Global Partnership on Artificial Intelligence are Australia, Canada, the European Union, France, Germany, India, Italy, Japan, Rep. Korea, Mexico, New Zealand, Singapore, Slovenia, the US and the UK. The GPAI Secretariat is hosted by the OECD in Paris, France. GPAI's mandate covers four themes, two of which are supported by the International Centre of Expertise in Montréal for the Advancement of Artificial Intelligence, namely, responsible AI and data governance. A corresponding centre of excellence in Paris, yet to be identified, will support the other two themes on the future of work and innovation, and commercialization. GPAI will also investigate how AI can be leveraged to respond to the COVID-19 pandemic.

UNESCO will be tabling an international instrument on the ethics of AI for adoption by 192 member states in November 2021.

Given the concerns about data exploitation, the European Union also developed an artificial intelligence policy, with a working group studying ways to assure confidence in the use of artificial intelligence. These were issued in two white papers in the midst of the COVID-19 pandemic. One of the policies on artificial intelligence is called A European Approach to Excellence and Trust.

From Fiction

 Not done This section is about fiction, and we only have room to cover the most popular tropes. This material below doesn't illustrate a major trope and places WP:UNDUE on this artist for this article (and is unsourced). Could not find a a place for this, as artificial intelligence in fiction has a very tight structure at this point and doesn't seem to be ready to accept discussion of random works.

In the 1980s, artist Hajime Sorayama's Sexy Robots series were painted and published in Japan depicting the actual organic human form with lifelike muscular metallic skins and later "the Gynoids" book followed that was used by or influenced movie makers including George Lucas and other creatives. Sorayama never considered these organic robots to be real part of nature but always an unnatural product of the human mind, a fantasy existing in the mind even when realized in actual form.

Citations needed for the material above

When the material above is moved into a sub article, we will need the citation it used. You should be able to find them here. Note that the citation format of the article was all over the map. ---- CharlesGillingham (talk) 09:02, 24 September 2021 (UTC)

Sources

• Luger, George; Stubblefield, William (2004). Artificial Intelligence: Structures and Strategies for Complex Problem Solving (5th ed.). Benjamin/Cummings. ISBN 978-0-8053-4780-7. Archived from the original on 26 July 2020. Retrieved 17 December 2019.
• Hochreiter, Sepp; Schmidhuber, Jürgen (1997), "Long Short-Term Memory", Neural Computation, 9 (8): 1735–1780, doi:10.1162/neco.1997.9.8.1735, PMID 9377276, S2CID 1915014
• Schmidhuber, J. (2015a). "Deep Learning in Neural Networks: An Overview". Neural Networks. 61: 85–117. arXiv:1404.7828. doi:10.1016/j.neunet.2014.09.003. PMID 25462637. S2CID 11715509.
• Howe, J. (November 1994). "Artificial Intelligence at Edinburgh University: a Perspective". Archived from the original on 15 May 2007. Retrieved 30 August 2007.
• Maker, Meg Houston (2006). "AI@50: AI Past, Present, Future". Dartmouth College. Archived from the original on 3 January 2007. Retrieved 16 October 2008.
• Domingos, Pedro (September 22, 2015). The Master Algorithm: How the Quest for the Ultimate Learning Machine Will Remake Our World. Basic Books. ISBN 978-0465065707.
• McGaughey, E (2018), Will Robots Automate Your Job Away? Full Employment, Basic Income, and Economic Democracy, p. SSRN part 2(3), SSRN 3044448 Archived 24 May 2018 at the Wayback Machine
• Schmidhuber, Jürgen (2015b). "Deep Learning". Scholarpedia. 10 (11): 32832. Bibcode:2015SchpJ..1032832S. doi:10.4249/scholarpedia.32832.
• Dechter, Rina (1986), Learning while searching in constraint-satisfaction problems, University of California, Computer Science Department, Cognitive Systems Laboratory Archived 19 April 2016 at the Wayback Machine
• Aizenberg, Igor; Aizenberg, Naum N.; Vandewalle, Joos P.L. (2000), Multi-Valued and Universal Binary Neurons: Theory, Learning and Applications., Springer Science & Business Media
• Ivakhnenko, Alexey (1965). Cybernetic Predicting Devices. Kiev: Naukova Dumka.</ref>
• Ivakhnenko, A. G. (1971). "Polynomial Theory of Complex Systems". IEEE Transactions on Systems, Man, and Cybernetics (4): 364–378. doi:10.1109/TSMC.1971.4308320. S2CID 17606980.
• Fukushima, K. (1980). "Neocognitron: A self-organizing neural network model for a mechanism of pattern recognition unaffected by shift in position". Biological Cybernetics. 36 (4): 193–202. doi:10.1007/bf00344251. PMID 7370364. S2CID 206775608.
• LeCun, Yann (2016), Slides on Deep LearningArchived 23 April 2016 at the Wayback Machine
• Silver, David; Schrittwieser, Julian; Simonyan, Karen; Antonoglou, Ioannis; Huang, Aja; Guez, Arthur; Hubert, Thomas; Baker, Lucas; Lai, Matthew; Bolton, Adrian; Chen, Yutian; Lillicrap, Timothy; Fan, Hui; Sifre, Laurent; Driessche, George van den; Graepel, Thore; Hassabis, Demis (19 October 2017). "Mastering the game of Go without human knowledge" (PDF). Nature. 550 (7676): 354–359. Bibcode:2017Natur.550..354S. doi:10.1038/nature24270. ISSN 0028-0836. PMID 29052630. S2CID 205261034. AlphaGo Lee... 12 convolutional layers
• Schmidhuber, J. (1992). "Learning complex, extended sequences using the principle of history compression". Neural Computation. 4 (2): 234–242. CiteSeerX 10.1.1.49.3934. doi:10.1162/neco.1992.4.2.234. S2CID 18271205.
• Graves, Alex; Fernandez, Santiago; Gomez, Faustino; Schmidhuber, Jürgen (2006), "Connectionist temporal classification: Labelling unsegmented sequence data with recurrent neural nets", Proceedings of ICML'06: 369–376, doi:10.1145/1143844.1143891, S2CID 9901844
• Hannun, Awni; Case, Carl; Casper, Jared; Catanzaro, Bryan; Diamos, Greg; Elsen, Erich; Prenger, Ryan; Satheesh, Sanjeev; Sengupta, Shubho; Coates, Adam; Ng, Andrew Y. (2014). "Deep Speech: Scaling up end-to-end speech recognition". arXiv:1412.5567 [cs.CL].
• Sak, Hasim; Senior, Andrew; Beaufays, Francoise (2014), "Long Short-Term Memory recurrent neural network architectures for large scale acoustic modeling", Proceedings of Interspeech 2014
• Li, Xiangang; Wu, Xihong (2015). "Constructing Long Short-Term Memory based Deep Recurrent Neural Networks for Large Vocabulary Speech Recognition". arXiv:1410.4281 [cs.CL].
• Sutskever, Ilya; Vinyals, Oriol; Le, Quoc V. (2014). "Sequence to Sequence Learning with Neural Networks". arXiv:1409.3215 [cs.CL].
• Jozefowicz, Rafal; Vinyals, Oriol; Schuster, Mike; Shazeer, Noam; Wu, Yonghui (2016). "Exploring the Limits of Language Modeling". arXiv:1602.02410 [cs.CL].
• Gillick, Dan; Brunk, Cliff; Vinyals, Oriol; Subramanya, Amarnag (2015). "Multilingual Language Processing From Bytes". arXiv:1512.00103 [cs.CL].
• Vinyals, Oriol; Toshev, Alexander; Bengio, Samy; Erhan, Dumitru (2015). "Show and Tell: A Neural Image Caption Generator". arXiv:1411.4555 [cs.CV].
• Hyötyniemi, Heikki (1996). "Turing machines are recurrent neural networks". Proceedings of STeP '96/Publications of the Finnish Artificial Intelligence Society: 13–24.
• Gödel, Kurt (1951). Some basic theorems on the foundations of mathematics and their implications. Gibbs Lecture. in Feferman, Solomon, ed. (1995). Kurt Gödel: Collected Works, Vol. III: Unpublished Essays and Lectures. Oxford University Press. pp. 304–23. ISBN 978-0-19-514722-3.
• Albus, J. S. (2002). "4-D/RCS: A Reference Model Architecture for Intelligent Unmanned Ground Vehicles" (PDF). In Gerhart, G.; Gunderson, R.; Shoemaker, C. (eds.). Proceedings of the SPIE AeroSense Session on Unmanned Ground Vehicle Technology. Unmanned Ground Vehicle Technology IV. Vol. 3693. pp. 11–20. Bibcode:2002SPIE.4715..303A. CiteSeerX 10.1.1.15.14. doi:10.1117/12.474462. S2CID 63339739. Archived from the original (PDF) on 25 July 2004.
• White Paper: On Artificial Intelligence – A European approach to excellence and trust (PDF). Brussels: European Commission. 2020. Archived (PDF) from the original on 20 February 2020. Retrieved 20 February 2020.
• Clark, Jack (8 December 2015b). "Why 2015 Was a Breakthrough Year in Artificial Intelligence". Bloomberg.com. Archived from the original on 23 November 2016. Retrieved 23 November 2016.
• Russell, Stuart J.; Norvig, Peter (2009). Artificial Intelligence: A Modern Approach (3rd ed.). Upper Saddle River, New Jersey: Prentice Hall. ISBN 978-0-13-604259-4..
• Weng, J.; McClelland; Pentland, A.; Sporns, O.; Stockman, I.; Sur, M.; Thelen, E. (2001). "Autonomous mental development by robots and animals" (PDF). Science. 291 (5504): 599–600. doi:10.1126/science.291.5504.599. PMID 11229402. S2CID 54131797. Archived (PDF) from the original on 4 September 2013. Retrieved 4 June 2013 – via msu.edu.
• Lungarella, M.; Metta, G.; Pfeifer, R.; Sandini, G. (2003). "Developmental robotics: a survey". Connection Science. 15 (4): 151–190. CiteSeerX 10.1.1.83.7615. doi:10.1080/09540090310001655110. S2CID 1452734.
• Asada, M.; Hosoda, K.; Kuniyoshi, Y.; Ishiguro, H.; Inui, T.; Yoshikawa, Y.; Ogino, M.; Yoshida, C. (2009). "Cognitive developmental robotics: a survey". IEEE Transactions on Autonomous Mental Development. 1 (1): 12–34. doi:10.1109/tamd.2009.2021702. S2CID 10168773.
• Oudeyer, P-Y. (2010). "On the impact of robotics in behavioral and cognitive sciences: from insect navigation to human cognitive development" (PDF). IEEE Transactions on Autonomous Mental Development. 2 (1): 2–16. doi:10.1109/tamd.2009.2039057. S2CID 6362217. Archived (PDF) from the original on 3 October 2018. Retrieved 4 June 2013.
• Russell, Stuart J.; Norvig, Peter (2003), Artificial Intelligence: A Modern Approach (2nd ed.), Upper Saddle River, New Jersey: Prentice Hall, ISBN 0-13-790395-2.
• McCarthy, John; Minsky, Marvin; Rochester, Nathan; Shannon, Claude (1955). "A Proposal for the Dartmouth Summer Research Project on Artificial Intelligence". Archived from the original on 26 August 2007. Retrieved 30 August 2007.
• Crevier, Daniel (1993). AI: The Tumultuous Search for Artificial Intelligence. New York, NY: BasicBooks. ISBN 0-465-02997-3..
• Kurzweil, Ray (2005). The Singularity is Near. Penguin Books. ISBN 978-0-670-03384-3.
• Ford, Martin; Colvin, Geoff (6 September 2015). "Will robots create more jobs than they destroy?". The Guardian. Archived from the original on 16 June 2018. Retrieved 13 January 2018.
• Russell, Stuart (October 8, 2019). Human Compatible: Artificial Intelligence and the Problem of Control. United States: Viking. ISBN 978-0-525-55861-3. OCLC 1083694322.
• Weizenbaum, Joseph (1976). Computer Power and Human Reason. San Francisco: W.H. Freeman & Company. ISBN 978-0-7167-0464-5.
• McCorduck, Pamela (2004), Machines Who Think (2nd ed.), Natick, MA: A. K. Peters, Ltd., ISBN 1-56881-205-1.
• Moravec, Hans (1988). Mind Children. Harvard University Press. ISBN 978-0-674-57616-2. Archived from the original on 26 July 2020. Retrieved 18 November 2019.
• Nilsson, Nils (1998). Artificial Intelligence: A New Synthesis. Morgan Kaufmann. ISBN 978-1-55860-467-4. Archived from the original on 26 July 2020. Retrieved 18 November 2019.
• Lucas, John (1961). "Minds, Machines and Gödel". In Anderson, A.R. (ed.). Minds and Machines. Archived from the original on 19 August 2007. Retrieved 30 August 2007.
• Clark, Jack (1 July 2015a). "Musk-Backed Group Probes Risks Behind Artificial Intelligence". Bloomberg.com. Archived from the original on 30 October 2015. Retrieved 30 October 2015.
• Hawkins, Jeff; Blakeslee, Sandra (2005). On Intelligence. New York: Owl Books. ISBN 978-0-8050-7853-4.
• Penrose, Roger (1989). The Emperor's New Mind: Concerning Computer, Minds and The Laws of Physics. Oxford University Press. ISBN 978-0-19-851973-7.
• Hofstadter, Douglas (1979). Gödel, Escher, Bach: an Eternal Golden Braid. New York: Vintage Books. ISBN 978-0-394-74502-2.
• UNESCO Science Report: the Race Against Time for Smarter Development. Paris: UNESCO. 11 June 2021. ISBN 978-92-3-100450-6.
• Turing, Alan (October 1950), "Computing Machinery and Intelligence", Mind, LIX (236): 433–460, doi:10.1093/mind/LIX.236.433, ISSN 0026-4423.

Unused citations from the article (not needed above)

 Not done These citations were not used in the article. Some of these could be "further reading", I suppose.

• Sikos, Leslie F. (June 2017). Description Logics in Multimedia Reasoning. Cham: Springer. doi:10.1007/978-3-319-54066-5. ISBN 978-3-319-54066-5. S2CID 3180114. Archived from the original on 29 August 2017.
• Jackson, Philip (1985). Introduction to Artificial Intelligence (2nd ed.). Dover. ISBN 978-0-486-24864-6. Archived from the original on 26 July 2020. Retrieved 4 March 2020.
• Neapolitan, Richard; Jiang, Xia (2018). Artificial Intelligence: With an Introduction to Machine Learning. Chapman & Hall/CRC. ISBN 978-1-138-50238-3. Archived from the original on 22 August 2020. Retrieved 3 January 2018.
• Winston, Patrick Henry (1984). Artificial Intelligence. Reading, MA: Addison-Wesley. ISBN 978-0-201-08259-3. Archived from the original on 26 July 2020. Retrieved 22 August 2020.
• Rich, Elaine (1983). Artificial Intelligence. McGraw-Hill. ISBN 978-0-07-052261-9. Archived from the original on 26 July 2020. Retrieved 17 December 2019.
• Bundy, Alan (1980). Artificial Intelligence: An Introductory Course (2nd ed.). Edinburgh University Press. ISBN 978-0-85224-410-4.
• Poole, David; Mackworth, Alan (2017). Artificial Intelligence: Foundations of Computational Agents (2nd ed.). Cambridge University Press. ISBN 978-1-107-19539-4. Archived from the original on 7 December 2017. Retrieved 6 December 2017.
• Auffarth, Ben (2020). Artificial Intelligence with Python Cookbook: Proven recipes for applying AI algorithms and deep learning techniques using TensorFlow 2.x and PyTorch 1.6 (1st ed.). Packt Publishing. ISBN 978-1-78913-396-7. Archived from the original on 8 January 2021. Retrieved 13 January 2021.
• Bach, Joscha (2008). Wang, Pei; Goertzel, Ben; Franklin, Stan (eds.). Seven Principles of Synthetic Intelligence. IOS Press. pp. 63–74. ISBN 978-1-58603-833-5. Archived from the original on 8 July 2016. Retrieved 16 February 2016. {{cite encyclopedia}}: |work= ignored (help)
• Brooks, R. A. (1991). "How to build complete creatures rather than isolated cognitive simulators". In VanLehn, K. (ed.). Architectures for Intelligence. Hillsdale, NJ: Lawrence Erlbaum Associates. pp. 225–239. CiteSeerX 10.1.1.52.9510.
• Buchanan, Bruce G. (2005). "A (Very) Brief History of Artificial Intelligence" (PDF). AI Magazine: 53–60. Archived from the original (PDF) on 26 September 2007.
• Diamond, David (December 2003). "The Love Machine; Building computers that care". Wired. Archived from the original on 18 May 2008.
• Dowe, D. L.; Hajek, A. R. (1997). "A computational extension to the Turing Test". Proceedings of the 4th Conference of the Australasian Cognitive Science Society. Archived from the original on 28 June 2011.
• Dreyfus, Hubert (1992). What Computers Still Can't Do. New York: MIT Press. ISBN 978-0-262-54067-4.
• Edelman, Gerald (23 November 2007). "Gerald Edelman – Neural Darwinism and Brain-based Devices". Talking Robots. Archived from the original on 8 October 2009.
• Gladwell, Malcolm (2005). Blink. New York: Little, Brown and Co. ISBN 978-0-316-17232-5.
• Goodman, Joanna (2016). Robots in Law: How Artificial Intelligence is Transforming Legal Services (1st ed.). Ark Group. ISBN 978-1-78358-264-8. Archived from the original on 8 November 2016. Retrieved 7 November 2016.
• Hernandez-Orallo, Jose (2000). "Beyond the Turing Test". Journal of Logic, Language and Information. 9 (4): 447–466. doi:10.1023/A:1008367325700. S2CID 14481982.
• Hernandez-Orallo, J.; Dowe, D. L. (2010). "Measuring Universal Intelligence: Towards an Anytime Intelligence Test". Artificial Intelligence. 174 (18): 1508–1539. CiteSeerX 10.1.1.295.9079. doi:10.1016/j.artint.2010.09.006.
• Hinton, G. E. (2007). "Learning multiple layers of representation". Trends in Cognitive Sciences. 11 (10): 428–434. doi:10.1016/j.tics.2007.09.004. PMID 17921042. S2CID 15066318.
• Holland, John H. (1975). Adaptation in Natural and Artificial Systems. University of Michigan Press. ISBN 978-0-262-58111-0. Archived from the original on 26 July 2020. Retrieved 17 December 2019.
• Kleine-Cosack, Christian (October 2006). "Recognition and Simulation of Emotions" (PDF). Archived from the original (PDF) on 28 May 2008.
• Hutter, M. (2012). "One Decade of Universal Artificial Intelligence". Theoretical Foundations of Artificial General Intelligence. Atlantis Thinking Machines. Vol. 4. pp. 67–88. CiteSeerX 10.1.1.228.8725. doi:10.2991/978-94-91216-62-6_5. ISBN 978-94-91216-61-9. S2CID 8888091.
• James, William (1884). "What is Emotion". Mind. 9 (34): 188–205. doi:10.1093/mind/os-IX.34.188.
• Kaplan, Andreas; Haenlein, Michael (2019). "Siri, Siri in my Hand, who's the Fairest in the Land? On the Interpretations, Illustrations and Implications of Artificial Intelligence". Business Horizons. 62: 15–25. doi:10.1016/j.bushor.2018.08.004.
• Koza, John R. (1992). Genetic Programming (On the Programming of Computers by Means of Natural Selection). MIT Press. Bibcode:1992gppc.book.....K. ISBN 978-0-262-11170-6.
• Kolata, G. (1982). "How can computers get common sense?". Science. 217 (4566): 1237–1238. Bibcode:1982Sci...217.1237K. doi:10.1126/science.217.4566.1237. PMID 17837639.
• Kumar, Gulshan; Kumar, Krishan (2012). "The Use of Artificial-Intelligence-Based Ensembles for Intrusion Detection: A Review". Applied Computational Intelligence and Soft Computing. 2012: 1–20. doi:10.1155/2012/850160.
• Kurzweil, Ray (1990), The Age of Intelligent Machines, Cambridge, MA: MIT Press, ISBN 0-262-11121-7
• Kurzweil, Ray (1999). The Age of Spiritual Machines. Penguin Books. ISBN 978-0-670-88217-5.
• Lakoff, George; Núñez, Rafael E. (2000). Where Mathematics Comes From: How the Embodied Mind Brings Mathematics into Being. Basic Books. ISBN 978-0-465-03771-1.
• Law, Diane (June 1994). Searle, Subsymbolic Functionalism and Synthetic Intelligence (Technical report). University of Texas at Austin. p. AI94-222. CiteSeerX 10.1.1.38.8384.
• McCarthy, John; Hayes, P. J. (1969). "Some philosophical problems from the standpoint of artificial intelligence". Machine Intelligence. 4: 463–502. CiteSeerX 10.1.1.85.5082. Archived from the original on 10 August 2007. Retrieved 30 August 2007.
• McCarthy, John (12 November 2007). "What Is Artificial Intelligence?". www-formal.stanford.edu. Archived from the original on 18 November 2015.
• McCarthy, John (2007-11-12). "Applications of AI". www-formal.stanford.edu. Archived from the original on 28 August 2016. Retrieved 25 September 2016.
• Minsky, Marvin (2006). The Emotion Machine. New York: Simon & Schusterl. ISBN 978-0-7432-7663-4.
• Norvig, Peter (25 June 2012). "On Chomsky and the Two Cultures of Statistical Learning". Peter Norvig. Archived from the original on 19 October 2014.
• Needham, Joseph (1986). Science and Civilization in China: Volume 2. Caves Books Ltd.
• Picard, Rosalind (1995). Affective Computing (PDF) (Technical report). MIT. 321. {{cite tech report}}: Unknown parameter |lay-source= ignored (help); Unknown parameter |lay-url= ignored (help)
• Poli, R.; Langdon, W. B.; McPhee, N. F. (2008). A Field Guide to Genetic Programming. Lulu.com. ISBN 978-1-4092-0073-4. Archived from the original on 8 August 2015. Retrieved 21 April 2008 – via gp-field-guide.org.uk.
• Rajani, Sandeep (2011). "Artificial Intelligence – Man or Machine" (PDF). International Journal of Information Technology and Knowledge Management. 4 (1): 173–176. Archived from the original (PDF) on 18 January 2013.
• Ronald, E. M. A. and Sipper, M. Intelligence is not enough: On the socialization of talking machines, Minds and Machines Archived 25 July 2020 at the Wayback Machine, vol. 11, no. 4, pp. 567–576, November 2001.
• Ronald, E. M. A. and Sipper, M. What use is a Turing chatterbox? Archived 25 July 2020 at the Wayback Machine, Communications of the ACM, vol. 43, no. 10, pp. 21–23, October 2000.
• "Science". August 1982. Archived from the original on 25 July 2020. Retrieved 16 February 2016.
• Shapiro, Stuart C. (1992). "Artificial Intelligence". In Shapiro, Stuart C. (ed.). Encyclopedia of Artificial Intelligence (PDF) (2nd ed.). New York: John Wiley. pp. 54–57. ISBN 978-0-471-50306-4. Archived (PDF) from the original on 1 February 2016. Retrieved 29 May 2009.
• van der Walt, Christiaan; Bernard, Etienne (2006). "Data characteristics that determine classifier performance" (PDF). Archived from the original (PDF) on 25 March 2009. Retrieved 5 August 2009.
• O'Brien, James; Marakas, George (2011). Management Information Systems (10th ed.). McGraw-Hill/Irwin. ISBN 978-0-07-337681-3.
• O'Connor, Kathleen Malone (1994). The alchemical creation of life (takwin) and other concepts of Genesis in medieval Islam (Dissertation). University of Pennsylvania. pp. 1–435. AAI9503804. Archived from the original on 5 December 2019. Retrieved 27 August 2008 – via Dissertations available from ProQuest.
• Skillings, Jonathan (3 July 2006). "Getting Machines to Think Like Us". cnet. Archived from the original on 16 November 2011. Retrieved 3 February 2011.