The future of AI and CogSci depends on a continued supply of able and enthusiatic young people studying the subjects at university and being interested in moving into research. An important prerequisite for this is encouraging children and young people to take study the appropriate subjects at school and to consider research as an exciting and promising career. Interest in studying AI at university has fallen in recent years and any researcher interested in the future of AI should be concerned that this trend does not continue. There are many ways in which researchers can help. There are many funding opportunities for researchers to get involved with or create outreach projects and many opportunities for them to get involved in schools, in science festivals and in other events where they can promote their own research as a way of stimulating young people and educate about the field in general. This page includes links that might be helpful for researchers with an interest in getting more involved in this. For further details, please contact (insert my committee address).
How to get involved
- Researchers in Residence: this scheme offers you the opportunity to use your expertise as a researcher and develop your communication skills by going into a school or college and helping create quality sceince experiences for the pupils. All RinRs are trained in how to communicate their research to non-specialists and undergo a Criminal Records check.
- Science and Engineering Ambassadors The Science and Engineering Ambassadors (SEAs) Programme is SETNET$B!G(Bs flagship programme. Ambassadors are individuals from a wide variety of STEM backgrounds, from all across the UK, who offer their time, enthusiasm and expertise to help schools inspire young people. Typical activities in which Ambassadors may get involved are: * supporting schools activities such as Science and Engineering clubs * helping with school's STEM competitions, events and awards * assisting in extra-curricular STEM experiences * offering mentoring, careers guidance and positive role models * helping to provide work-based placements for teachers and students.
- Postgrad and Postdocs can get involved in public engagement and the social and ethical implications of their research by entering a poster in BA perspectives
- The Meet the scientist project holds events at science centres which enables researchers to engage in dialogue with the general public.
- FameLab is a national competition to find the UK's best new talent in sceince communication. Win a masterclass in science communication, the change to appear on Channel 4 and a cash prize of £2000.
- The Nuffield foundation offers bursaries for post-16 science students to work alongside practising scientists
- You can also apply for media and communication skills training courses that are run by RCUK.
- You can get funding of up to £2000 to support a public engagement project during Science week, which is held throughout the country.
- If you want to get involved to a deeper level, you can apply for an EPRSC Partnership for Public Engagement grant.
Artificial Intelligence is a broad discipline. It contains many diverse subfields and it has strong links to areas such as Cognitive Science and Philosphy of Mind. The term "Artificial Intelligence" was first coined by Prof. John McCarthy for a Conference on the subject held at Dartmouth in 1956. McCarthy defines the subject as the "science and engineering of making intelligent machines, especially intelligent computer programs". The term "intelligence" is, of course, the subject of both scientific and philosophical debate. Within Artificial Intelligence many researchers use the term to mean giving computers behaviours which would be thought intelligent in human beings.
Artificial Intelligence research covers a wide range of topics. It includes making computers better at tasks that are widely perceived as intelligent, such as proving mathematical theorems. It also investigates processes we have only recently recognised are of significant difficulty, such as recognising objects in pictures. It even investigates problems which had not even been considered before the invention of computers, such as making it easier to find information on the world wide web.
The field of Artificial Intelligence does not solely concern itself with replicating "human-like" intelligence. Artificial Intelligence systems, such as chess-playing programs, do not necessarily work in "human-like" ways. There is a whole separate field, of Cognitive Science, which devotes itself to understanding the ways humans (and indeed other animals) think. Clearly there are many fertile links between the two fields. In many cases a better understanding of human cognition can lead to advances in Artificial Intelligence. At the same time finding a way to get a computer to peform a task can shed light on ways that humans might think.
Artificial Intelligence, unsurprisingly, has been the inspiration for much speculative fiction. This often involves stories in which computers and/or robots behave like particularly intelligent and physically strong versions of humans. Similar scenarios are occasionally raised in the Press. These tend to be presented in an alarmist fashion. There are a number of both practical and philosophical problems with these ideas, certainly in the short to medium term. It is also entirely unclear that the nature of Artificial Intelligence research is likely to produce this kind of intelligent and malevolent robot. As discussed above, much Artificial Intelligence research is focused on making computers easier for humans to use in various ways. The AISB doesn't deny these possibilities. However we feel that the attention they attract often draws scrutiny away from more immediate concerns. Such concerns include ethically dubious uses of computers here and now and the degree to which our society is already dependent on them.
The fields of Artificial Intelligence and Cognitive Science are wide and fascinating parts of Computer Science and Psychology. The Artificial Intelligence challenge includes the ability to perceive, learn, store information, reason about what is known, communicate using human language, and interact with the physical environment, e.g., move objects. All of these abilities are still being investigated by individual researchers and none can be considered "solved". This makes Artificial Intelligence and Cognitive Science fascinating and exciting topics of study.
General Artificial Intelligence and Cognitive Science Resources
We do not maintain a comprehensive list of Artificial Intelligence and Cognitive Science web resources. However this is a short list of web sites that we feel represent our subject in an accurate fashion.
What is AI? - This site is maintained by John McCarthy, one of the founders of the field, and he keeps it up-to-date. It is a good place to start exploring the nature of Artificial Intelligence.
AAAI AI Topics - The Association for the Advancement of Artificial Intelligence maintains a large website full of resources. This particular page is aimed at young people and provides material they might find useful in school projects as well as suggestions for reference material.
In Our Time: Artificial Intelligence. A BBC Radio 4 broadcast from December 2005, available via Listen Again from the BBC Website.
Libraries and Collections
University of Edinburgh Informatics Collection
The University of Edinburgh's Informatics Collection is housed in the Main Library, George Square and includes approximately 140 current journal subscriptions and some 3,500 books plus access for University staff and students to numerous online resources. Information about admission for visitors can be found under "Membership, Admission, Registration and Access"
Information about AI resources, especially with reference to the 2002 fire which destroyed the University's AI Library is available from Library Services following the AI Library fire.
The Informatics Digitisation Project arose as a result of the destruction of the AI Collection. Existing Open Access: Research reports, preprints and postprints are also available online.
Grand Challenges in Computer Research
The UKCRC sponsors the formation of Grand Challenges in Computer Research.
"The chief purpose of the formulation and promulgation of a grand challenge is the advancement of science. A grand challenge represents a commitment by a significant scientific community to work together towards a common goal, agreed to be valuable and achievable within a predicted timescale. The challenge is formulated by the scientists themselves as a focus for the research that they wish to pursue in any case. It is independent of any political initiatives or prior allocation of special funding. It may involve a thousand man-years of research effort, drawn from many countries and spread over ten years or more."
The following is the list of current Grand Challenge web sites. We feel that there is a place of Artificial Intelligence and Cognitive Science Research within all these challenges. However GC5, The Architecture of Brain and Mind, is of most obvious interest to AISB members. We are also aware of significant Artificial Intelligence involvement in both GC6, Dependable Systems Evolution and GC8, Learning for Life.
GC1: In Vivo - In Silico
GC2/4: Ubiquitous Computing: Experience, Design and Science
GC3: Memories for Life
GC5: The Architecture of Brain and Mind
GC6: Dependable Systems Evolution
GC7: Journeys in Nonclassical Computation
GC8: Learning for Life
Degrees in Artificial Intelligence
It is impractical to list all UK degrees with significant Artificial Intelligence components. However the British Council maintains a good list.
The following links are to respectable degree course web sites.
Suitable keywords when searching for relevant degrees on these sites are Artificial Intelligence; Cognitive Science and Robotics
UCAS general course search
Hobson's (Postgraduate) - more information information available about some courses.
Education UK (run by the British Council)
The following is an incomplete list of AI related organisation.
The Association for the advancement Artificial Intelligence
The European Coordinating Committee for Artificial Intelligence.
Natural Computing Applications Forum
YouTube video of the Lighthill debate from 1973
In 1973, Professor Sir James Lighthill was asked by Parliament to evaluate the state of AI research in the United Kingdom. His report, now called the Lighthill report, criticized the utter failure of AI to achieve its "grandiose objectives." He concluded that nothing being done in AI couldn't be done in other sciences. He specifically mentioned the problem of "combinatorial explosion" or "intractability", which implied that many of AI's most successful algorithms would grind to a halt on real world problems and were only suitable for solving "toy" versions.
The report was contested in a debate broadcast in the BBC "Controversy" series in 1973. The debate "The general purpose robot is a mirage" from the Royal Institute was Lighthill versus the team of Michie, McCarthy and Gregory.
The report led to the near-complete dismantling of AI research in England.
Starring: James Lighthill, Donald Michie, Richard Gregory and John McCarthy.
Activities and Funding for Researchers (EPSRC).
Intediciplinary Collaboration (NESTA) - parts of this cover PU activities.
Beacons for Public Engagement (RCUK) - list of university based collaborative centres.
Communication Skills and Media Training Courses (Royal Society) - EPSRC will pay for these courses as part of a normal science grant.
Funding for Public Engagement (Science and Technology Facilities Council).
Public Engagement (Wellcome).
About this Page
We have constructed this page in order to sketch how AI is linked to various other disciplines, in both the sciences and humanities. We have done this not only in the hope of helping students and others who are just starting to study AI but also of facilitating further interactions between the AI community and other communities.
The set of links is not exhaustive, nor is the explanation given for each individual link.
In particular, we do not include all disciplines that do or could use AI tools and/or contribute tools to AI. We concentrate rather on disciplines where there is profound interaction in terms of research ideas. In most cases the interaction is two-way.
We also omit discussion of the somewhat contentious topic of the relationship of AI to Computer Science, which may be treated in future in a separate document. (Most people would agree that the boundary between AI and CS is fuzzy and subjective. But some people think AI is part of CS, some people think the reverse is true, and some people think they merely overlap. Some people think AI is often too vague to be counted as computer "science", other people think that CS is scientifically deficient if it doesn't overlap or include AI.)
AI is intrinsically one source of creative suggestions for Psychology. Much AI research is directed ultimately at proposing detailed models for explaining the cognitive, perceptual and emotional qualities of people and animals. AI is not just about creating useful artefacts that have abilities grossly comparable to those of people and animals, though this is itself of immense importance.
Conversely, Psychological research can help AI:
Psychology can suggest cognitive, perceptual, emotional, ... mechanisms for use in AI systems even when there is no intention for those systems to act as models for biological creatures.
AI systems interacting with people, especially via humanoid avatars (screen images) or robotic bodies, need often to at least look as though they have appropriate emotions, even if they don't actually have them. The psychology of how emotion is linked to thought, perception and body movements therefore comes into play for this practical reason.
More subtly:- Imagine an AI system that is meant to be able to communicate with ordinary people about everyday matters. The system needs to be able to know how people commonsensically reason about things (or needs to be able simulate such reasoning). This in order to be able to understand people's linguistic and other communications about those things and in order, conversely, to be able to communicate things that will be understandable to people. Thus, results from Psychology about how people think about everyday matters are useful. This is so even if the AI agent is not itself meant to be a model of a person.
(Neuro)Physiology and Biology more generally
These disciplines can help AI:
Results from the study of the nature and working of biological bodies and especially of brains are helpful in the development of computational infrastructures for AI (e.g.: artificial neural networks; algorithmic systems analogous to the immune system) and robotic hardware and body control systems.
The study of biological and cultural evolution has led to artefactual evolutionary computing, with applications within AI ranging from those at a fine scale (e.g., self-developing algorithms for doing some specific cognitive task) to those at a gross scale, as in evolution of populations of whole AI agents.
The complex computation effected by large swarms of relatively simple organisms (e.g., ants) has led to analogous forms of computation within AI, again both within an individual agent and in terms of large groups of whole agents.
AI can help those disciplines: Development of thinking agents and robots within AI, even if it has no specific intention to illuminate biology, can in fact do so, in terms of physiology, evolution, swarm activity, etc.
AI has much overlap with Philosophy, at both a grand scale and a more detailed technical scale.
AI is intrinsically concerned with grand Philosophical issues such as the fundamental nature of thought, consciousness, mind/body relationship, morality, free will, aesthetic qualities, value, time, etc., and more specialized philosophical issues such as the fundamental nature of language, representation, information, logic, computation, rule-following, symbols and causation.
Many detailed technical problems studied in Philosophy are also technical challenges for AI. One example is the thorny representational and cognitive issues raised by propositional attitude reports (roughly: sentences reporting mental states of agents) and more generally by the need for AI systems to represent and reason about mental states of agents. Other examples include vague language and information, counterfactual hypotheses, obligations, and the nature of information about past and future.
AI raises new problems for Philosophy or intensifies the urgency of old problems. E.g., the fluidity of intelligent software agents on the web, where we can imagine agents merging, splitting or cloning themselves, or not having clear boundaries in the first place, raises/intensifies Philosophical issues about personhood and the notion of a self.
AI can provide working models embodying Philosophical theories and thus serve as a testbed for them (testing for their consistency, completeness, economy of structure, etc.)
On the other hand, Philosophical research on issues important to AI can have a level of refined subtlety and sophistication, and can have a breadth or degree of ambitiousness, that most AI researchers do not have the resources or opportunity to achieve.
Note: Logic could be deemed covered by Philosophy above, and is not so often counted as a separate discipline; but it has an especially salient role for AI. Also of interest for AI is the Psychological issue of what sorts of logical processing people do or don't do, get right or mess up, but this is covered more by the interaction of AI with Psychology.
AI has much overlap and interaction with Logic. Developments in Logic fundamentally affect not only research into explicitly logic-based processing in AI but also much other work on forms of representation, depending on the ways in which those forms translate (or fail to translate) into logic. But also the needs of AI to deal with taxonomies (and "ontologies"), uncertainty, inconsistent information, vagueness, time, mental states, obligations, etc. lead to new or revised logics being invented and explored that are of interest in the field of Logic itself. One type of case is variations on modal logic for mental states where the usual provisions are augmented with considerations of mental resource limitations; another is the development of description logic (to couch taxonomies, ontologies, etc.).
Results such as Goedel's about the limitations of formal logic have a general importance for computational theory, and of course bear in particular upon the theoretical capabilities of intelligent agents based on logic. However, a more special point that affects AI is the use of those limitations in arguments that human beings are fundamentally superior to any conceivable (machine-like) artefact. Such arguments attack the very idea of "strong" AI (that genuine thought, consciousness, etc. can be artefactually realized). There is of course much contention over the validity of the arguments.
Linguistics (incl. study of sign language and gesture)
Note: The fields of Pragmatics and Rhetoric are considered to be covered in this section.
One of the main needs of an AI agent interacting with people is to be able to communicate with people in natural language (i.e., human language ... and the more "natural" in the ordinary sense the better!) and/or to be able to understand natural language communications between people. Linguistics is therefore an important source of knowledge and ideas in the development of such agents.
Conversely, AI can contribute ideas to Linguistics:
Practical approaches taken in the engineering of AI artefacts that have some capability in natural language processing may stimulate ideas and fruitful controversies about the nature of human language itself. E.g., statistical approaches to machine translation of languages can act as a foil to traditional theorizing about the nature of the human language faculty.
The practical imperative in AI to deal with some of the most abstruse problems in language such as discourse structure, point of view expression, metaphor and emotional expression, even when only mundane conversation or text is at issue, leads to theoretical models that can be of use to Linguistics in its own studies of these matters.
Study of the automatic development of language within populations of AI agents contributes to fundamental understanding of language itself.
Communication in sign language and through gesture (especially as linked to spoken language) has in recent years exploded in importance both within Linguistics and within AI and Computer Science.
Studies of gesture have suggested the nature of some of the cognitive processes underlying spoken language itself. This is useful for AI.
Art, Music, Literature ... and Creativity
("Art" below includes cinema, music, dance, literature, etc. as well visual art in the usual sense. Also the field of Aesthetics is intended to be covered by this section.)
Insofar as AI is intended to emulate people and illuminate their nature it must attend not just to "dry" forms of cognition such as everyday reasoning and expert problem solving but also to the human artistic capabilities. Indeed, insofar as artistic endeavour appears central to the nature of people it would be dangerous within AI (and Psychology, Philosophy, etc.) to neglect it. Also, aesthetic sensibility is important even in science, given the tendency to prefer "elegant" theories and models.
More generally, the study of what creativity is and how to achieve it artefactually is important for AI, and goes beyond art into scientific theorizing, expert problem solving and commonsense reasoning. One technical topic shared between artistic creativity and non-artistic forms of it, and that has seen much detailed attention in AI, Psychology and elsewhere is the use of analogy.
Art is centrally concerned with human emotion, a topic that is of increasing importance within AI (and Computer Science). See also the comments on Psychology above.
Issues of representation and point of view that arise in artistic forms of expression present cognitively revealing problems beyond the more classical issues arising in language, logic, etc.
These problems also link to non-art issues in AI. For instance, the way point of view and mode of consciousness are expressed in language is much studied in literary theory and is important also for ordinary language processing. The way paintings (whether photographically realistic or abstract) express and represent things is linked to the way diagrams do; and the study of diagrammatic representation/reasoning is an important topic for AI.
AI intensifies and refines biological, evolutionary and philosophical questions of whether the arts and creativity are uniquely human qualities or, going in one direction, are shared with sub-human creatures or by constrast, in the other direction, require something mystical beyond physical processes.
Economics, Politics, Sociology, Business Management, etc.
AI is much concerned with the study of communities of agents, not just with single agents. Thus, many issues of cooperation, conflict, trust, social norms, governance, resource usage, organizational dynamics, etc. are shared with disciplines such as economics, politics, sociology and business management.
AI agents operating within real markets (e.g., doing automated stock trading) are becoming a crucial part of economic reality and may have fundamental effects on economic theory insofar as this theory needs to be concerned with timescales of and motivations for decisions.
The ability of AI agents to help in Medicine, Law, Policing, etc. raises important issues of ethics, responsibility, accountability and statutory regulation.
Virtual worlds not only have their own internal economics, politics, etc. but also interact with or overlap with the real world. This is a trend that is set to rise rapidly. The use of AI within virtual worlds will increasingly intensify the problems for humankind that this trend raises.
Medicine / Psychiatry / Healthcare / ...
The development of AI agents capable of performing or helping with medical diagnosis, treatment design, psychiatric therapy, more informal lifestyle advising, elder care, nursing, clinical trial analysis etc. has been an interest since the early days of AI and in some respects is currently burgeoning. Aside from possible practical AI contributions are the ethical, regulatory, etc. issues raised for theoreticians and practitioners in Medicine, etc., as mentioned above.
There is no real converse in the direction from Medicine etc. to AI (aside from the point that any application area of AI can of course stimulate the development of important new ideas in AI). One day, though, AI agents will need some analogue of human healthcare!
AI and Law have a strong shared interest in the nature of commonsense reasoning, expert reasoning, evidence, persuasion, deception, exploration (investigation), intention, etc. Law is also an important application area for AI expert systems.
The use of examples within case-based reasoning in AI (a form of analogy-based reasoning, and similar also to "example-based reasoning" is Psychology) is highly analogous to the use of prior legal cases in Law. The relationship of case-based reasoning to "rule-based" reasoning in AI is analogous to the relationship between reasoning via cases and reasoning via statutes in Law.
The increase over time of the level of intelligence and degree of embeddedness in society of AI agents will lead to new challenges for Law, on matters such as responsibility, intention, evidence, ownership (notably of intellectual property, of web-crawling software agents, especially if they can collaborate to create other such agents), punishment, and the fundamental nature of legal processes themselves.