Artificial General Intelligence

Last update : August 7, 2013

Artificial General Intelligence (AGI) is an emerging research field aiming at the building of thinking machines; that is, general-purpose systems with intelligence comparable to that of the human mind (and perhaps ultimately well beyond human general intelligence). While this was the original goal of Artificial Intelligence (AI), the mainstream of AI research has turned toward domain-dependent and problem-specific solutions; therefore it has become necessary to use a new name to indicate research that still pursues the Grand AI Dream. Similar labels for this kind of research include Strong AI, Human-level AI, etc. Other AI researchers prefer the term of Synthetic Intelligence.

The research on AGI is interdisciplinary, focused on whole systems and includes scientific and philosophical investigation and software engineering.

Artificial General Intelligence Research Institute

The term AGI was first used by Mark Avrum Gubrud in November 1997. Fifty years after the launch of the Artificial Intelligence Project in Dartmouth in 1956, Ben Goertzel, Phil Goetz, Pei Wang and Bruce Klein organized the first Artificial General Intelligence Research Institute (AGIRI) workshop in May 2006 to bridge the gap between narrow AI and general-purpose AI. The AGI Research Institute was founded in 2001 with the mission to foster the creation of powerful and ethically positive Artificial General Intelligence. The institute is sponsored by Novamente LLC.

The aspects of Artificial General Intelligence are explained by Pei Wang and Ben Goertzel  in the introduction of their book Advances in Artificial General Intelligence (IOS Press, 2007).

The first conference on Artificial General Intelligence (AGI-08) was organized by AGIRI in March 2008 in Memphis, Tennessee, USA, in association with the Association for the Advancement of Artificial Intelligence (AAAI).

Artificial General Intelligence Society

Ben Goertzel, Pei Wang, Joscha Bach and others founded in September 2011 the Artificial General Intelligence Society (AGI society), a nonprofit organization with the following goals:

  • promote the study of artificial general intelligence (AGI), and the design of AGI systems
  • facilitate co-operation and communication among those interested in the study and pursuit of AGI
  • hold conferences and meetings for the communication of knowledge concerning AGI
  • produce publications regarding AGI research and development
  • publicize and disseminate by other means knowledge and views concerning AGI

The organization of the annual Artificial General Intelligence conference series, which was started in 2008 by AGIRI, has been taken over by the AGI society. The next conference (AGI-2013) will be held in Beijing, China, July 31 – August 3, 2013.

Some additional informations about AGI are available at the following links :

More links are provided in the updated post about Artificial Intelligence.

Supertoy Teddy and Huggable

Supertoy Teddy

Supertoy Teddy

Supertoy Teddy is the world’s first talking teddy with a mind of its own and the ability to hold real conversations with those who speak to it. It has been developed by Ashley Conlan (CEO of Supertoy Robotics) and Kartsen Fluegge (CEO of Pannous GmbH), the creators of the successful app Jeannie, the Siri style chatbot that has been downloaded over 3 million times on mobile devices.

Supertoy Teddy uses artificial intelligence (AI). A smartphone acts as its brawn and the internet server as its brain. Supertoy Teddy’s robotic mouth moves in synchronization to what it says and inbuilt speakers enhance the volume of its voice. Role play will be added to the Supertoy Teddy and several costumes and dresses will be sold at the online shop.

The robot’s hardware is simple: just an audio in/out interface and a motor for mouth animation. Supertoy Teddy connects via standard audio plug to an iOS or Android device. Asley Conlan suggests putting the phone inside the Supertoy for realism. The magic is in the software, which has evolved from the popular Jeannie chatbot app.

[jwplayer player=”1″ mediaid=”12984″]

The creators of Supertoy Teddy have started a Kickstarter campaign for crowdfunding during the funding period July 24, 2013 – August 23, 2013. I am one of the bakers of the project.

Huggable (MIT)

Huggable (MIT)

Huggable is a similar type of robotic companion that has being developed at the MIT Media Lab (Personal Robots Group) for healthcare, education, and social communication applications a few years ago. The early technical development of the Huggable was supported in part by a Microsoft iCampus grant in 2006.

Compressing Human Knowledge

Marcus Hutter, a German computer scientist and professor at the Australian National University, funded in August 2006 a 50.000 euros cash prize, which rewards data compression improvements on the first 100.000.000 characters of a specific version of English Wikipedia (envik8). Specifically, the prize awards 500 euros for each percent improvement in the compressed size of the file enwik8. The prize baseline was 18,324,887 bytes, achieved by PAQ8F, a free lossless data compression archiver. The contest is open ended and is open to everyone. The ongoing competition is organized by Marcus Hutter, Matt Mahoney and Jim Bowery.

The goal of the Hutter Prize is to encourage research in artificial intelligence (AI). The organizers believe that text compression and AI are equivalent problems. There is no general solution to achieve this goal because the descriptive complexity is not computable. In algorithmic information theory the measure of the computational resources needed to specify an object is called the Kolmogorov complexity, named after Andrey Nikolaevich Kolmogorov, a Soviet mathematician.


Last update : August 6, 2013

Supercomputers were introduced in the 1960s and were designed primarily by Seymour Cray at Control Data Corporation (CDC), and later at Cray Research. While the supercomputers of the 1970s used only a few processors, in the 1990s, machines with thousands of processors began to appear and by the end of the 20th century, massively parallel supercomputers with tens of thousands of “off-the-shelf” processors were the norm.

ChipTest, Deep Thought and Deep Blue supercomputers

ChipTest, Deep Thought and Deep Blue were chess computers. The chess project was started at Carnegie Mellon University by Feng-hsiung Hsu in 1985. He and his collaborators were hired by IBM Research in 1989 to continue their work to build a chess machine that could defeat the world champion. On May 11, 1997, Deep Blue, with human intervention between games, won the second six-game match against world champion Garry Kasparov by two wins to one with three draws.

Blue Gene supercomputers

Blue Gene is an IBM project aimed at designing supercomputers that can reach operating speeds in the petaFLOPS range, with low power consumption. The initial design for Blue Gene was based on an early version of the Cyclops64 architecture, designed by Monty Denneau. The project created three generations of supercomputers, Blue Gene/L, Blue Gene/P, and Blue Gene/Q. In 2004, the first IBM Blue Gene computer became the fastest supercomputer in the world.

Watson supercomputers

Watson is an artificial intelligence computer system capable of answering questions posed in natural language, developed in IBM’s DeepQA project by a research team led by principal investigator David Ferrucci. Watson was named after IBM’s first president, Thomas J. Watson. In 2011, as a test of its abilities, Watson competed on the quiz show Jeopardy!. Watson had access to 200 million pages of structured and unstructured content consuming four terabytes of disk storage including the full text of Wikipedia, but was not connected to the Internet during the game.

IBM describes Watson as “an application of advanced Natural Language Processing, Information Retrieval, Knowledge Representation and Reasoning, and Machine Learning technologies to the field of open domain question answering“.  IBM’s DeepQA technology is used for hypothesis generation, massive evidence gathering, analysis, and scoring.

Watson is related to Artificial Intelligence and to the research of commonsense knowledge, the collection of facts and information that an ordinary person is expected to know.

Artificial Intelligence

Last update : August 9, 2013

Artificial intelligence (AI) is the intelligence of machines and the branch of computer science which aims to create it.The term was coined by John McCarthy in 1955. The field of AI research was founded at a conference on the campus of Dartmouth College in the summer of 1956. The attendees, including John McCarthy, Marvin Minsky, Allen Newell and Herbert A. Simon, became the leaders of Artificial Intelligence research for many decades.

Good Old-Fashioned Artificial Intelligence (GOFAI)

AI research began in the mid 1950s after the Dartmouth conference. The field AI was founded on the claim that a central property of humans, intelligence, can be so precisely described that it can be simulated by a machine. The first generation of AI researchers were convinced that this sort of AI was possible and that it would exist in just a few decades.

In the early 1970s, it became obvious that researchers had grossly underestimated the difficulty of the project. By the 1990s, AI researchers had gained a reputation for making promises they could not keep. The AI research suffered from longstanding differences of opinion how it should be done and from the application of widely differeing tools.

The field of AI regressed into a multitude of relatively well insulated domains like logics, neural learning, expert systems, chatbots, robotics, semantic web, case based reasoning etc., each with their own goals and methodologies. These subfields, which often failed to communicate with each other, are often referred as applied AI, narrow AI or weak AI.

The old original approach to achieving artificial intelligence is called GOFAI. The term was coined by John Haugeland in his 1986 book Artificial Intelligence: The Very Idea.

Weak Artificial intelligence

After the AI winter, the mainstream of AI research has turned with success toward domain-dependent and problem-specific solutions. These subfields of weak AI have grown up around particular institutions and individual researchers, some of them are listed hereafter:

Peter Norvig, Google’s head of research, and Eric Horvitz, a distinguished scientist at Microsoft Research, are optimistic about the future of machine intelligence. They spoke recently to an audience at the Computer History Museum in Palo Alto, California, about the promise of AI. Afterward, they talked with Technology Review‘s IT editor, Tom Simonite.

A few AI searchers continue to believe that artificial intelligence could match or exceed human intelligence. The term strong AI, now in wide use, was introduced for this category of AI by the philosopher John Searle of the University of California at Berkeley. Among his notable concepts is the Chinese Room, a thought experiment which is an argument against strong AI.

Strong Artificial Intelligence

Strong AI is the intelligence of a machine that could successfully perform any intellectual task that a human being can. Strong AI is associated with traits such as consciousness, sentience, sapience (wisdom) and self-awareness observed in living beings.

There is a wide agreement among AI researchers that strong artificial intelligence is required to do the following :

  • reason, use strategy, solve puzzles and make judgements under uncertainty
  • represent knowledge, including commonsense knowledge
  • plan
  • learn
  • communicate in natural language
  • integrate all these skills towards common goals

Other important capabilities include the ability to sense (see, …) and the ability to act (move and manipulate objects, …) in the observed world.

Some AI researchers adopted the term of Artificial General Intelligence (AGI) to refer to the advanced interdisciplinary research field of strong AI. Other AI researchers prefer the term of Synthetic Intelligence to make a clear distinction with GOFAI.

The following links provide some informations about the history and the concepts of Artificial Intelligence :

A list of organizations and institutions dealing with Artificial Intelligence is shown below :

“Artificial intelligence is no match for human stupidity!”

Semantic Web

Last Update : October 7, 2012

The Semantic Web is a collaborative movement led by the international standards body W3C. The Semantic Web is a Web of Data, as opposed to the existing Web of Documents. The goal of the Web of Data is to enable computers to do more useful work and to develop systems that can support trusted interactions over the network.

The Web of Data is empowered by new technologies such as RDFa (Resource Description Framework-in attributes), SPARQL, OWL (Web Ontology Language), SKOS (Simple Knowledge Organization System), Microdata and Open Graph.

HTML (HyperText Markup Language) remains still the main markup language for displaying web pages and other information that can be displayed in a web browser.

Semantic HTML refers to the semantic elements and attributes of HTML (h1, h2, …, p, …), as opposed to the presentational HTML elements and attributes (center, font, b, …). The acronym POSH was coined in 2007 for semantic HTML, as a shorthand abbreviation for “plain old semantic HTML”.

HTML5 introduced a few new structural elements :

  • <header> : this tag replaces the <div class=”header”>, commonly used in the past by most designers. The header element contains introductory information to a section or page.
  • <footer> : same as above, it’s the well known <div class=”footer”>. The footer element is for marking up the baseline of the current page and of each section contained in the page.
  • <nav> : replacement for <div class=”navigation”>. The nav element is reserved for the primary navigation. Not all link groups in a page or section need to be contained within the <nav> element.
  • <section> : this is the replacement for the generic flow container <div> when it contains related content. <div> is a block-level element with no additional semantic meaning, whereas <section> is a sectioning element which has normally a header and a footer and represents a generic document or application section.
  • <article> : the <article> element represents a portion of a page or section which can stand alone and makes sense even outside the context of the page. Like <section>, an <article> generally has a header and a footer. You should avoid nesting an <article> inside another <article>.

HTML5 tag <aside>

  • <aside> : this tag is used to represent content that is related to the surrounding content within an section, article or web page, but could still stand alone in its own right. (see figure at right). This type of content is often represented in sidebars.
  • <hgroup> : A special header element that must contain at least two <h1>-<h6> tags and nothing else. It’s a group of titles with subtitles. Make sure to maintain the <h1> – <h6> hierarchy.

RDFa is a W3C Recommendation that adds a set of attribute-level extensions (rich metadata) to web documents. RDFa 1.1 was approved in June 2012. It differs from RDFa 1.0 in that it no longer relies on the XML-specific namespace mechanism, but ca be used with non-XML document types such as HTML 4 or HTML 5. eRDF is an alternative to RDFa. SPARQL is an RDF query language. On 15 January 2008, SPARQL 1.0 became an official W3C Recommendation. OWL is a family of knowledge representation languages for authoring ontologies. An ontology formally represents knowledge as a set of concepts within a domain in computer science and information science, and the relationships among those concepts. Ontologies are the structural frameworks for organizing information and are used, among others, in artificial intelligence. SKOS is a family of formal languages designed for representation of of structured controlled vocabulary (thesauri, classification schemes, taxonomies, …). Microdata is a WHATWG specification used to nest semantics within existing content on web pages. The Open Graph protocol, originally created by Facebook, enables any web page to become a rich object in a social graph.

All these technologies help computers such as search engines and web crawlers better understand what information is contained in a web page, providing better search results for users.

Another set of simple, structured open data formats, built upon existing standards, is Microformats. One difference with the other semantic technologies is that Microformats is designed for humans first and machines second.

The following list provides links to some useful blogs and tutorials about the semantic web:

AIML (Artificial Intelligence Markup Language)

Last update : July 17, 2013

AIML virtual assistant

Virtual Assistant Denise by Guile 3D Studio

AIML (Artificial Intelligence Markup Language) is an XML-compliant language that’s easy to learn, and makes it possible to customize an artificial intelligence chat robot or creating one from scratch within minutes. AIML is free open-source software provided by the ALICE A.I. Foundation, a non-profit research and training organization. ALICE stands for Artificial Linguistic Internet Computer Entity, an award-winning free natural language artificial intelligence chat robot. The AIML 2.0 draft specification was released on January 16, 2013.

A AIML beginners guide is available at the website of the ALICE  A.I. Foundation.

There are various AIML sets available, among them the free Annotated ALICE AIML set, a revised release of the scripts comprising the award winning chat robot ALICE. The ALICE A. I. Foundation is also offering the commercial version Superbot 2.1 (999 US$) that helps you to create a totally unique custom bot personality for your web site or application.

The other free softwares available are implementations of the ALICE chatbot engine in different computer languages, tools and knowledge bases. Documentation, specifications, tutorials and showcases are available at the website. Chatting with an original ALICEBOT is possible in the Hall of Fame of Digital Art at Leslie’s Artgallery.

Another more advanced implementation of an AIML robot is Denise, a virtual assistant software created by Guile 3D Studio.

An outstanding tool to program a robot brain in AIML is the free GaitoBot AIML editor provided by the german company Springwald Software.

LOVOTICS = Love + Robotics

Last update : May 18, 2013


The Human and robot may fall in love one day!  That was the goal of Hooman Samani when he was an AI (Artficial Intelligence) researcher at the Keio-NUS CUTE Center, which is a collaborative Social Robots Lab between National University of Singapore and Keio University of Japan. In 2013, Dr. Hooman Samani is Director of the AIART Lab (Artificial Intelligence and Robotics Technology Laboratory) and an Assistant Professor at the Department of Electrical Engineering, College of Electrical Engineering and Computer Science, National Taipei University, Taiwan. He is the founder of Lovotics research and developed several applications in the field of human-to-robot relationship.

Lovotics includes Artificial Endocrine System, Probabilistic Love Assembly and Affective State Transition modules. They are based on physiology of love, psychology of love and emotional models respectively. Artificial Endocrine System produces artificial hormones such as Dopamine, Oxytocin, Serotonin and Endorphin, imitating the human endocrine system.

Affective State Transition controls various affective parameters of love, for example facial expressions, voice, gesture. They allow Lovotics robot to reason about other person’s emotions and state of mind.

Lovotics is a multidisciplinary research field utilizing fundamentals concepts from robotics, artificial intelligence, philosophy, psychology, biology, anthropology, neuroscience, social science, computer science and engineering. Kissenger, Mini-Surrogate and XOXO are applications of Lovotics.


Last update : August 6, 2013

Verbot 5

Verbots (Verbally Enchanted Software Robots)  is a popular chatterbot program and Artificial Intelligence Software Development Kit (SDK) for the Windows platform and for the web, created by Dr. Michael Mauldin and Peter Plantec.

Some milestones of the history of verbots are presented hereafter :

  • 1989 : TinyMUD Gloria
  • 1990 : TinyMUD Julia
  • 1991 : participation of Julia in the first Loebner Prize contest
  • 1994 : chatterbot Julia
  • 1997 : creation of Virtual Personalities, Inc.
  • 2000 : production release of the virtual human interface Sylvie
  • 2004 : release of the Verbot 4 version
  • 2006 : start of Verbots Online
  • 2010 : relase of the Verbot 5 version

Version 4 of verbots was based on MS Agent which has been discontinued by Microsoft in Windows 7.  A properties viewer of MS agents has been created by AbhiSoft Technologies. A related scripting software has been developed by the same company who also provides a file repository for MS Agents.

Version 5 of verbots uses characters made up of 22 SAPI5 viseme groups and animations. The Conversive Character Studio Application allows you to easily create your own talking characters that are compatible with Verbots and VerbotsOnline using high-quality SAPI 5 tts voices. Conversive characters are defined in a .css file in xml format. Sample visemes are available at the verbots wiki website. Animations are a collection of frames which are displayed on the screen in sequence.

The different verbots file types are :

  • ckb : Compiled KnowledgeBase
  • csv : Comma Separated Values
  • vkb : Verbot KnowledgeBase
  • vrp : Verbot Replacement Profile
  • vsn : Verbot Synonyms

The templates to create the Verbots brain are the following :

  • My answers
  • My Knowledge Bases
  • My Design
  • Install

In the Online version you can browse the chat logs, manage your account and list your bot in the online directory. Several tags are available to commande the Verbot. A Verbot editor allows to create and edit the different templates. KnowledgeBases are created from a collection of Rules. Rules contain Inputs and Outputs. Rules can be Primary Rules, Child Rules or Virtual Child Rules. Conditionals, variables and regular expressions are further means to set up a personality. Special inputs allow to start and stop animations, embedded C# code modules allows to execute programs, schedue tags allow to trigger time events, commands are used to open web adresses or to run applications.

The Teaching.vkb KnowledgeBase allows new rules to be dynamically added while chatting.

ChatVerbots for IRC and AIM are available as beta versions.

The following tutorials about Verbots are available :

  • Creating personalities
  • Creating your first rule
  • Creating child rules
  • Knowledge Base templates and csv files

Communities discussing about verbots are listed below :

Concerning  AIML (Artificial Intelligence Markup Language maintained by the Alice Foundation ), verbots don’t comply to this standard. Verbots KB (Knowledge Base) and AIML both are XML based, but the format is different, the working of engine is different, the usability is different.

A free tool to convert AIML files to Verbots KB files is available at the Verbots website.

The Alice Foundation is more active, more dynamic and more professional, compared to the Verbots Community, whereas the Verbots Technology offers some outstanding features.

The Verbots Online Service was closing down at the end of August 2012. Free webhosting for AIML is still available  : Pandorabots for AIML. To fully customize your Bot and to give him a Voice (TTS), a paid subscription is required. SitePal.

People researching on Artificial Consciousness

Based mainly on the outstanding informations at the website, an updated list of people researching in the field of machine consciousness is shown below :

  • Dr. Raúl Arrabales Moreno : Assistant Professor, Computer Science Department, Computer Science and Artificial Intelligence, Universidad Carlos III de Madrid
  • Dr. Igor Aleksander : Emeritus Professor of Neural Systems Engineering in the Department of Electrical and Electronic Engineering at Imperial College London, UK; Fellow of the Royal Academy of Engineering
  • Dr Will Browne : Senior Lecturer, School of Engineering and Computer Science, Victoria University of Wellington, New Zealand
  • Dr. Antonio Chella : Head of RoboticsLab, Dipartimento di Ingegneria Informatica (DINFO), Università di Palermo, Italy
  • Dr. Ron Chrisley : Reader in Philosophy, Director, Center for Research in Cognitive Science, University of Sussex, UK
  • Dr. Axel Cleeremans : Research Director, National Fund for Scientific Research, Member of the Royal Academy of Belgium, Consciousness, Cognition & Computation Group, Université Libre de Bruxelles CP 191, Belgium
  • Dr. Stan Franklin : W. Harry Feinstone Interdisciplinary Research Professor, Cognitive Computing Research Group
    Department of Computer Science, Institute for Intelligent Systems, The University of Memphis, USA
  • Dr. David Gamez : Research Associate, Department of Computing, Imperial College, London, UK
  • Dr. Ben Goertzel : Cross-disciplinary scientist, engineer, entrepreneur, manager, writer, speaker; CTO, Genescient Corp, Irvine CA, USA; CEO and Chief Scientist, Novamente LLC, Rockville MD, USA; CEO and Chief Scientist, Biomind LLC, Rockville MD, USA
  • Steve Grand : Director, Cyberlife Research Ltd., Somerset, UK
  • Dr. Pentti O A Haikonen : Adjunct Professor, Department of Philosophy, University of Illinois at Springfield, USA
  • Owen Holland : Professor of cognitive robotics (Informatics) in the Sackler Centre for Consciousness Science at the University of Sussex, UK
  • Dr. Ray Kurzweil : entreprenuer, leading inventor, author, restless genius, ultimate thinking machine; he has received nineteen honorary Doctorates and honors from three U.S. presidents; in 2002 he was inducted into the National Inventor’s Hall of Fame in USA
  • Dr. Riccardo Manzotti : Assistant Professor in Psychology, IULM University, Milan, Italy
  • Dr. Hugo Gravato Marques : Artificial Intelligence Laboratory, Department of Informatics, University of Zurich, Switzerland
  • Dr. Michael Loren Mauldin (alias Fuzzy) :  Founder and chief scientist of Lycos ; Director of Conversive, Inc.
  • Peter Plantec : Clinical psychologist, animator, virtual human designer ; author of the book Virtual Humans; founder of Virtual Personalities, Inc. (now Conversive, Inc.) in order to create the first virtual human interface Vperson (now Verbots)
  • Dr. Uma Ramamurthy : Asst. Professor & Director of Research Informatics, Dan L. Duncan Institute for Clinical and Translational Research, Baylor College of Medicine, Houston, USA
  • Dr. Ricardo Sanz : Professor in Systems Engineering and Automatic Control and coordinator of the Autonomous Systems Laboratory research group at the Universidad Politécnica de Madrid in Madrid, Spain
  • Dr. Anil Seth : Co-Director, Sackler Centre for Consciousness Science (SCCS), University of Sussex; Reader, School of Informatics, University of Sussex; EPSRC Leadership Fellow; Visiting Professor, Dept of Psychology, University of Amsterdam
  • Dr. Murray Patrick Shanahan : Professor of Cognitive Robotics, Department of Computing, Computational Neurodynamics Group, Imperial College London, UK
  • Dr. Aaron Sloman : School of Computer Science, The University of Birmingham, UK