A fundamental difficulty in artificial intelligence is that nobody really knows what intelligence is, especially for systems with senses, environments, motivations and cognitive capacities which are very different to our own.

Although there is no strict consensus among experts over the definition of intelligence for humans, most definitions share many key features. In all cases, intelligence is a property of an entity, which we will call the agent, that interacts with an external problem or situation, which we will call the environment. An agent’s intelligence is typically related to its ability to succeed with respect to one or more objectives, which we will call the goal. The emphasis on learning, adaptation and flexibility common to many definitions implies that the environment is not fully known to the agent. Thus true intelligence requires the ability to deal with a wide range of possibilities, not just a few specific situations. Putting these things together gives us our informal definition: Intelligence measures an agent’s general ability to achieve goals in a wide range of environments. We are confident that this definition captures the essence of many common perspectives on intelligence. It also describes what we would like to achieve in machines: A very general capacity to adapt and perform well in a wide range of situations.

{ Shane Legg and Marcus Hutter | Continue reading | PDF }

Artificial general intelligence (AGI) refers to research aimed at tackling the full problem of artificial intelligence, that is, create truly intelligent agents. This sets it apart from most AI research which aims at solving relatively narrow domains, such as character recognition, motion planning, or increasing player satisfaction in games. But how do we know when an agent is truly intelligent? A common point of reference in the AGI community is Legg and Hutter’s formal definition of universal intelligence, which has the appeal of simplicity and generality but is unfortunately incomputable. (…)

Intelligence is one of those interesting concepts that everyone has an opinion about, but few people are able to give a definition for – and when they do, their definitions tend to disagree with each other. And curiously, the consensus opinions change over time: consider for example a number of indicators for human intelligence like arithmetic skills, memory capacity, chess playing, theorem proving – all of which were commonly employed in the past, but since machines now outperform humans on those tasks, they have fallen into disuse. We refer the interested reader to a comprehensive treatment of the subject matter in Legg (2008).

The current artificial intelligence literature features a panoply of benchmarks, many of which, unfortunately, are very narrow, applicable only on a small class of tasks. This is not to say that they cannot be useful for advancing the field, but in retrospect it often becomes clear how little an advance on a narrow task contributed to the general field. For example, researchers used to argue that serious progress on a game as complex as chess would necessarily generate many insights, and the techniques employed in the solution would be useful for real-world problems – well, no. (…)

Legg and Hutter propose their definition as a basis for any test of artificial general intelligence. Among the advantages they list are its wide range of applicability (from random to super-human), its objectivity, its universality, and the fact that it is formally defined.

Unfortunately however, it suffers from two major limitations: a) Incomputability: Universal intelligence is incomputable, because the Kolmogorov complexity is incomputable for any environment (due to the halting problem). b) Unlimited resources: The authors deliberately do not include any consideration of time or space resources in their definition. This means that two agents that act identically in theory will be assigned the exact same intelligence Υ, even if one of them requires infinitely more computational resources to choose its action (i.e. would never get to do any action in practice) than the other.

{ Tom Schaul, Julian Togelius, Jürgen Schmidhuber, Measuring Intelligence through Games, 2011 | Continue reading | PDF }

{ George Widener, for future robots, orientation map of events, for the year 4421| Elenore Weber of Ricco/Maresca Gallery in New York defined Widener, one of the most widely collected living outsider artists, as “a high-functioning savant” and a “lightning calculator.” His bio reads: “Like some people with aspergers, he is gifted in dates, numbers, and drawing. In his memory, he has several thousand historical dates, thousands of calendars, and more than a thousand census statistics.” | Salon }

Long before we communicated with language, we communicated with our bodies, especially our faces. Everyone knows we ‘talk’ with facial expressions, but do we ‘hear’ ourselves with them?

{ Thoughts on Thoughts | Continue reading }

The Turing test is a test of a machine’s ability to exhibit intelligent behavior. A human judge engages in a natural language conversation with one human and one machine, each emulating human responses. All participants are separated from one another. If the judge cannot reliably tell the machine from the human, the machine is said to have passed the test.

{ Wikipedia | Continue reading }

images { 1 | 2. Trisha Donnelly }

One group of Australian researchers have managed to teach robots to do something that, until now, was the reserve of humans and a few other animals: they’ve taught them how to invent and use spoken language. The robots, called LingoDroids, are introduced to each other. In order to share information, they need to communicate. Since they don’t share a common language, they do the next best thing: they make one up. The LingoDroids invent words to describe areas on their maps, speak the word aloud to the other robot, and then find a way to connect the word and the place, the same way a human would point to themselves and speak their name to someone who doesn’t speak their language.”

{ Slashdot | Continue reading }

artwork { Thomas Schütte, United Enemies, 1994-95 | fimo, fabric, wood glass and PVC }

{ Poker Bots Invade Online Gambling }

Each year for the past two decades, the artificial-intelligence community has convened for the field’s most anticipated and controversial event—a meeting to confer the Loebner Prize on the winner of a competition called the Turing Test. The test is named for the British mathematician Alan Turing, one of the founders of computer science, who in 1950 attempted to answer one of the field’s earliest questions: can machines think? That is, would it ever be possible to construct a computer so sophisticated that it could actually be said to be thinking, to be intelligent, to have a mind? And if indeed there were, someday, such a machine: how would we know?

Instead of debating this question on purely theoretical grounds, Turing proposed an experiment. Several judges each pose questions, via computer terminal, to several pairs of unseen correspondents, one a human “confederate,” the other a computer program, and attempt to discern which is which. The dialogue can range from small talk to trivia questions, from celebrity gossip to heavy-duty philosophy—the whole gamut of human conversation. Turing predicted that by the year 2000, computers would be able to fool 30 percent of human judges after five minutes of conversation, and that as a result, one would “be able to speak of machines thinking without expecting to be contradicted.”

Turing’s prediction has not come to pass; however, at the 2008 contest, the top-scoring computer program missed that mark by just a single vote.

{ The Atlantic | Continue reading }

related { Developed over four years at an estimated cost of more than $30 million, IBM’s “Jeopardy”-playing computer, Watson, will face the quiz show’s grand masters, Ken Jennings and Brad Rutter, in two games to be aired Feb. 14, 15 and 16. Doubts remain about how well Watson can process the endless subtleties of human language. WSJ | full story }

photo { Brian Ulrich }

A new program enables a robot to detect whether another robot is susceptible to lies, and to use its gullibility against it by telling lies, researchers claim.

The robot could be capable of deceiving humans in a similar way, according to the scientists, based at the Georgia Institute of Technology.

{ The Guardian | Continue reading }

Scientists are trying to teach robots to read - so they can understand road signs and shop names to ‘live’ for themselves.

Experts believe developing literate artificial intelligence should be relatively simple because computers are already able to turn scanned books into text.

{ Daily Mail | Continue reading }

photo { Mark King }

Reverse-engineering the human brain so we can simulate it using computers may be just two decades away, says Ray Kurzweil, artificial intelligence expert and author of the best-selling book The Singularity is Near.

It would be the first step toward creating machines that are more powerful than the human brain. These supercomputers could be networked into a cloud computing architecture to amplify their processing capabilities. Meanwhile, algorithms that power them could get more intelligent. Together these could create the ultimate machine that can help us handle the challenges of the future, says Kurzweil.

{ Wired | Continue reading }

related { The Man Who Made a Copy of Himself. A Japanese roboticist is building androids to understand humans–starting with himself. | ieee Spectrum | full story }

{ Four hexagonal robots, each with a propeller at its center, hook up using magnets to create one stable flying machine. }

{ Spring/Summer 1999, Mcqueen ended his show w/ model Shalom Harlow standin’ in a white dress on a rotatin’ platform, bein’ spray painted by robotic arms. }

Mirage: The Omnidroid 9000 is a top-secret military fighting robot. Artificial intelligence allows it to solve any problem it’s presented with, and, unfortunately…

Mr. Incredible: Let me guess. It became smart enough to wonder why it had to take orders.

Mirage: We lost control, and now it’s loose in the jungle, threatening our facility.

{ The Incredibles, 2004 }

Dave: Open the pod bay doors, HAL.

HAL: I’m sorry, Dave. I’m afraid I can’t do that.

Dave: What’s the problem?

HAL: I think you know what the problem is just as well as I do.

Dave: I don’t know what you’re talking about, HAL.

HAL: I know that you and Frank were planning to disconnect me, and I’m afraid that’s something I cannot allow to happen.

{ 2001: A Space Odyssey, 1968 }

related { Mechanized cooks invade the kitchen | NY Times | Full story }