I.B.M. scientists previously devised a chess-playing program to run on a supercomputer called Deep Blue. That program beat the world champion Garry Kasparov in a controversial 1997 match (Mr. Kasparov called the match unfair and secured a draw in a later one against another version of the program).

But chess is a game of limits, with pieces that have clearly defined powers. “Jeopardy!” requires a program with the suppleness to weigh an almost infinite range of relationships and to make subtle comparisons and interpretations. The software must interact with humans on their own terms, and fast.

Indeed, the creators of the system — which the company refers to as Watson, after the I.B.M. founder, Thomas J. Watson Sr. — said they were not yet confident their system would be able to compete successfully on the show, on which human champions typically provide correct responses 85 percent of the time.

“The big goal is to get computers to be able to converse in human terms,” said the team leader, David A. Ferrucci, an I.B.M. artificial intelligence researcher. “And we’re not there yet.”

The team is aiming not at a true thinking machine but at a new class of software that can “understand” human questions and respond to them correctly. Such a program would have enormous economic implications.

Despite more than four decades of experimentation in artificial intelligence, scientists have made only modest progress until now toward building machines that can understand language and interact with humans.

The proposed contest is an effort by I.B.M. to prove that its researchers can make significant technical progress by picking “grand challenges” like its early chess foray. The new bid is based on three years of work by a team that has grown to 20 experts in fields like natural language processing, machine learning and information retrieval.

Under the rules of the match that the company has negotiated with the “Jeopardy!” producers, the computer will not have to emulate all human qualities. It will receive questions as electronic text. The human contestants will both see the text of each question and hear it spoken by the show’s host, Alex Trebek.

The computer will respond with a synthesized voice to answer questions and to choose follow-up categories. I.B.M. researchers said they planned to move a Blue Gene supercomputer to Los Angeles for the contest. To approximate the dimensions of the challenge faced by the human contestants, the computer will not be connected to the Internet, but will make its answers based on text that it has “read,” or processed and indexed, before the show.

There is some skepticism among researchers in the field about the effort. “To me it seems more like a demonstration than a grand challenge,” said Peter Norvig, a computer scientist who is director of research at Google. “This will explore lots of different capabilities, but it won’t change the way the field works.”

The I.B.M. researchers and “Jeopardy!” producers said they were considering what form their cybercontestant would take and what gender it would assume. One possibility would be to use an animated avatar that would appear on a computer display.

“We’ve only begun to talk about it,” said Harry Friedman, the executive producer of “Jeopardy!” “We all agree that it shouldn’t look like Robby the Robot.”

Mr. Friedman added that they were also thinking about whom the human contestants should be and were considering inviting Ken Jennings, the “Jeopardy!” contestant who won 74 consecutive times and collected $2.52 million in 2004.

I.B.M. will not reveal precisely how large the system’s internal database would be. The actual amount of information could be a significant fraction of the Web now indexed by Google, but artificial intelligence researchers said that having access to more information would not be the most significant key to improving the system’s performance.

Eric Nyberg, a computer scientist at Carnegie Mellon University, is collaborating with I.B.M. on research to devise computing systems capable of answering questions that are not limited to specific topics. The real difficulty, Dr. Nyberg said, is not searching a database but getting the computer to understand what it should be searching for.

The system must be able to deal with analogies, puns, double entendres and relationships like size and location, all at lightning speed.

In a demonstration match here at the I.B.M. laboratory against two researchers recently, Watson appeared to be both aggressive and competent, but also made the occasional puzzling blunder.

For example, given the statement, “Bordered by Syria and Israel, this small country is only 135 miles long and 35 miles wide,” Watson beat its human competitors by quickly answering, “What is Lebanon?”

Moments later, however, the program stumbled when it decided it had high confidence that a “sheet” was a fruit.

The way to deal with such problems, Dr. Ferrucci said, is to improve the program’s ability to understand the way “Jeopardy!” clues are offered. The complexity of the challenge is underscored by the subtlety involved in capturing the exact meaning of a spoken sentence. For example, the sentence “I never said she stole my money” can have seven different meanings depending on which word is stressed.

“We love those sentences,” Dr. Nyberg said. “Those are the ones we talk about when we’re sitting around having beers after work.”

Source: New York Times

http://www.nytimes.com/2009/04/27/technology/27jeopardy.html

The ultimate goal, according to a recent request for research proposals, is to “elucidate core computational algorithms of the mind and brain.” Good luck with that, guys.

It’s one in a heap of different Office projects to try to teach machines to act more like living things. “Nature has used evolution to build materials and sensors that outperform current sensors (for example, a spider’s haircells can detect air flow at low levels even in a noisy background),” the Office writes. So it’s got a second program, to not only “mimic existing natural sensory systems, but also add existing capabilities to these organisms” so they can more “precise[ly] control” their God-given gifts.

For example, maybe the military can develop better “active and passive camouflage” by learning from creatures who are able to change color, to hide from their predators. Maybe the armed forces can improve on eznymes which would eat away at an enemy’s gear. Maybe the military can bioengineer the organisms living in extreme heat, or extreme acidity, to make our equipment stronger.

The Office also wants to know what makes collections of living creatures tick. So the Office is looking to assemble a “fundamental understanding of the interactions between demographic groups… to explain and predict outcomes between competing factions within geographic regions.” It wants to “identify and quantify cultural variability” to model the effects of an “info warfare campaign” online.

Once that’s done, it’s back to digitizing brainwork. “New computational and mathematical principles of cognition are needed to form a symbiosis between human and machine systems,” the Office says.

source: Wired

http://www.wired.com/dangerroom/2009/05/air-force-looks-for-core-algorithms-of-human-thought/

The study, published in Science, illustrates how nanoscale elements can organize themselves over a large sheet of semiconductor film. The researchers expect that when applied to electronic media, their discovery will improve the efficiency of data storage, savings which can then be transferred to improve other pieces of electronics besides just storage, like high-definition screens and solar cells.

Similar attempts have previously been made to improve data storage on semiconductor films, but have consistently failed because the polymers—which are known to link together, on their own, in precise patterns—lose their organized structure when the film being used increases in area, rendering them useless for storing memory.

Lead researchers Ting Xu from the University of California at Berkeley and Thomas Russell from the University of Massachusetts at Amherst overcame this by layering the film of block copolymers onto the surface of a commercially available sapphire crystal. When the crystal is cut at an angle—a common procedure known as a miscut—and heated to 1,300 to 1,500 degrees Centigrade (2,372 to 2,732 degrees Fahrenheit) for 24 hours, its surface reorganizes into a highly ordered pattern of sawtooth ridges that can then be used to guide the self-assembly of the block polymers [Science Daily].

With this technique, the only limit to the size of an array of block copolymers is the size of the sapphire, Xu said. Once a sapphire is heated up and the pattern is created, the template could be reused. Both the crystals and the polymer chains could be obtained commercially, Xu said [PC World]. The researchers say the technology could make nearly perfect arrays of semiconductor material that are about 15 times denser than anything achieved previously [Reuters].

Using the technology, it might also be possible to achieve a high-definition picture with 3-nanometer pixels, potentially as large as a stadium JumboTron, Xu said. Another possibility is more dense photovoltaic cells that capture the sun’s energy more efficiently….

The new technology could create chip features just 3nm [nanometers] across, far outstripping current microprocessor manufacturing techniques, which at their best create features about 45nm across [PC World].

Source: http://blogs.discovermagazine.com/80beats/2009/02/22/self-organizing-nanotech-could-store-250-dvds-on-one-coin-size-surface/

“What we have shown is that the bee brain, which contains less than 1 million neurons, is actually very good at learning to master complex tasks. Computer and imaging technology programmers who are working on solving complex visual recognition tasks using minimal hardware resources will find this research useful,” Dr Dyer said.

“Most current artificial intelligence (AI) recognition systems perform poorly at reliably recognising faces from different viewpoints. However the bees have shown they can recognise novel views of rotated faces using a mechanism of interpolating or image averaging previously learnt views.”

The findings show that despite the highly constrained neural resources of the insects (their brains are 0.01 per cent the size of the human brain) their ability has evolved so that they’re able to process complex visual recognition tasks.

The researchers individually trained different groups of free flying bees with a sugar reward for making correct choices, or alternatively the bees were punished with a bitter tasting solution for incorrect choices. Faces were presented on a vertical screen and bees slowly learnt to fly to the correct target faces. Over the course of a day a bee brain learned a complex task, and then when tested in non-rewarded tests (to totally excluded cues like olfaction) only bees that had experience multiple views (e.g. faces at both 0° and 60°) were able to solve a novel rotational angle of 30°.

Dr Dyer said the discovery helps to answer a fundamental question about how brains solve complex image rotational problems by either image averaging or mentally rotating previously learnt views.

“Bee brains clearly use image interpolation to solve the problem. In other words, bees that had learnt what a particular face looked like from two different viewpoints could then recognise a novel view of this target face. However, bees that had only learnt a single view could not recognise novel views,” Dr Dyer said.

The study, performed over two years in Australia and Germany by Dr Dyer with the support of the US Air Force Office of Scientific Research (AFOSR), and Dr Quoc Vuong from Newcastle University UK, was published in the science journal PLoS ONE.

“The relationships between different components of the object often dramatically change when viewed from different angles but it is amazing to find the bees’ brains have evolved clever mechanisms for problem solving which may help develop improved models for AI face recognition systems,” Dr Dyer said.

Source: http://www.sciencedaily.com/releases/2009/01/090123101211.htm

Not that it will necessarily have the same kind of consciousness as humans: it is unlikely to be wondering who it is, for instance. To Francis Heylighen, who studies consciousness and artificial intelligence at the Free University of Brussels (VUB) in Belgium, consciousness is merely a system of mechanisms for making information processing more efficient by adding a level of control over which of the brain’s processes get the most resources. “Adding consciousness is more a matter of fine-tuning and increasing control… than a jump to a wholly different level,” Heylighen says.

How might this manifest itself? Heylighen speculates that it might turn the internet into a self-aware network that constantly strives to become better at what it does, reorganising itself and filling gaps in its own knowledge and abilities.

If it is not already semiconscious, we could do various things to help wake it up, such as requiring the net to monitor its own knowledge gaps and do something about them. It shouldn’t be something to fear, says Goertzel: “The outlook for humanity is probably better in the case that an emergent, coherent and purposeful internet mind develops.”

Heylighen agrees, but warns that we might find it a little disappointing. “We probably would not notice a whole lot of a difference, initially,” he says.

And when might this begin? According to Heylighen, it all depends on internet fashion trends. If the effort that has gone into developing social networking sites goes into developing internet consciousness, it could happen within a decade, he says.

With more than 40 organizations in India hiring its services, CRL has made a mark in a field where only a few organizations have dared to venture. In a period of recession, Eka has opened up new possibilities for industries that require the massive crunching power of supercomputers, but only during certain times of their product development lifecycle.

For example, Tata Elxsi used Eka’s processing power to crunch the time required for rendering the animation movie, ‘Roadside Romeo.’ The activity, which would have taken the firm approximately 36-40 months to digitally render the movie in a studio, took only six months, due to the computing power of Eka. It is also significant to note that the firm achieved this feat using only one-third of the processing power of Eka.

Similarly, leading aerospace company, Boeing, is using Eka’s capability to bring its ideas faster to the market, by offering design and simulation support. Group company, Tata Motors, is using Eka for vehicle simulation and testing digital prototypes. CRL is also looking at potential opportunities in sectors such as Life Science, weather forecasting, animation and in the automotive and aviation industries. This is just a glimpse of the processing power.

Speaking with Seetha Rama Krishna, Head, HPC Engineering and Operations, CRL, one can gauge the ambitious goals of this small company. “‘Supercomputing made easy’ is our goal,” declares Seetha Rama Krishna, while elaborating on his firm’s vision to take supercomputers down to the retail level.

CRL’s initiative is a pioneering effort, as it is the first time that a corporate institution is taking the lead in extending the domain of High Performance Computing (HPC) from the academic field, to the enterprise. This is partly because a large-scale supercomputing infrastructure has been typically owned by government institutions, and is not largely used to its full potential. Being research focused, these institutions have little inclination or capability to deliver it as a utility service.

On the other hand, while private institutions in the oil and gas sector, or the automotive industry, would love to use a supercomputer, they cannot justify the cost of investing in a supercomputer that will be used only during specific periods. CRL is attempting to walk the tight rope between these two worlds, by offering services that are cost-effective even for small companies. “As an Indian company, it is in our ‘genes’ to be cost-effective, and deliver services that are at par or better than our global counterparts,” says Krishna, elaborating on the huge interest global firms have shown in his company’s services. Krishna certainly has the experience to walk the talk, as he has been the founder of the HPC solutions division at C-DAC, the state-owned firm that gave India its first supercomputer, way back in the 80’s.

CRL has succeeded because few organizations today have the financial bandwidth to afford a supercomputer; or because they are unwilling to invest and pay for the administration costs of maintaining a supercomputer. In this scenario, the concept of offering computing power as a service, has struck the right chord, as clients are happy to do this on a need basis at costs they can afford.

Eka is also attracting huge attention from Indian scientific and research institutions, as research shows that a majority of these institutions have supercomputers that are idle for a significant amount of time. Additionally, a significant number of institutions do not have resources with the requisite skills to effectively use or utilize a supercomputer. Eka is looking to occupy this space, with supercomputing solutions that can be effectively used by a large as well as small enterprise.

PAY-PER-USE
To encourage more enterprises to start using the supercomputers on rent, CRL is offering the services through three options: a pay-per-use model, a fixed capacity model, and through turnkey based customized solutions. Krishna envisages percolating the concept of supercomputers on rent, to small-scale enterprises or even professionals who might want to use the processing power of a supercomputer, for a specific period.

While few have succeeded in this field, the present economic condition is perhaps the perfect time to make this concept successful. “Eka’s tremendous number-crunching ability is ideal for research labs that require computing power for smaller durations, but do not have the money or technical skills to build or maintain these kind of solutions,” says Sandeep Lodha, Vice President, Netweb Technologies, a firm with huge domain experience in the HPC field. The firm has implemented over 60 HPC installations for some major Indian scientific institutions, and a number of firms in the bioinformatics domain, which are using supercomputers to cut down the time required for drug discovery. Netweb also has a small setup in its office where it offers supercomputing power on rent to some of its customers, who are contemplating buying their own setup.

Lodha says that once the cost-value equation becomes clear, more global organizations will come to India to outsource their data crunching requirements. A small trend in this direction is already taking place; Eka has been attracting huge interest from overseas clients interested in using the domain capabilities of the CRL team to test their applications.

“We are aiming to be a catalyst in this space and believe that as we push down the price, and offer it as a utility model, we can effectively remove the capacity constraint of users – physically and mentally. Imagine the impact, when enterprises are given the power to crunch design times, because of advanced computing capability. If more Indian enterprises have the ability to use this capability, one can deliver advanced products faster to the market, and lead the market,” says Krishna, speaking on the strategic impact that supercomputers can make, in generating insights from vast amounts of data, which would not be possible using traditional sources of computing.

By offering the supercomputing platform as a service, and bringing the benefits of a platform that has historically been out of reach for most enterprises, CRL has the potential to accelerate design and research, and revolutionize a new industry. Similar to what Tata Motors has done with the launch of Nano, can another Tata group company blaze a trail and create a new market? Watch this space!

Exponential growth in every industry eventually has to come to an end, according Anderson, who cited railroads and speed increases in the aircraft industry, the report said.

“A generation or two of continued exponential growth will likely continue only for leading-edge chips such as multicore microprocessors, but more designers are finding that everyday applications do not require the latest physical designs,” Anderson said in the EE Times’ report. Anderson also cited the staggering costs of research and fabs (factories) as a formidable barrier for continued advancement. Few companies can afford chip plants that typically cost billions of dollars to build and maintain.

So, what does the future hold? Anderson cited three technologies: optical interconnects, 3D chips–which have circuits and components stacked on top of each other–and accelerator-based processing as seeing significant advancements, the report said. The latter technology, accelerators, is hot right now.

In addition to IBM, companies such as Nvidia and Advanced Micro Devices’ ATI unit supply graphics-processor-based computers to accelerate scientific, engineering, and animation applications. Intel is also expected to bring out its Larrabee chip later this year or early next year that can be used as an accelerator.

Brooke Crothers is a former editor at large at CNET News.com, and has been an editor for the Asian weekly version of the Wall Street Journal. He writes for the CNET Blog Network, and is not a current employee of CNET. Contact him at mbcrothers@gmail.com. Disclosure.

Since it knew the precise positions of all the balls, it was very easy for it to pot the ball every time. This was fine for the highest level of difficulty, but for easy mode I simply gave the AI a random angular deviation to the shot.

Toward the end of the project, we got some feedback from the client that the AI was “too good.” I was puzzled by this and assumed the person wanted the expert mode to be slightly less accurate. So I changed that. But then I heard complaints about the decreased accuracy, and again that the AI was still too good.

Eventually the clients paid a visit to our offices and tried to demonstrate in person what they meant. It gradually came out that they thought the problem was actually with the “easy” mode.

They liked that the computer missed a lot of shots, but they thought that the positional play was too good. The computer always seemed to be leaving the white ball in a convenient position after its shot, either playing for safety or lining up another ball. They wanted that changed.

The problem was, there was no positional play! The eventual position of the white ball was actually completely random. The AI only calculated where the cue ball should hit the object ball in order to make that object ball go into a pocket.

It then blindly shot the cue ball toward that point with a speed proportional to the distance needed to travel, scaled by the angle, plus some fudge factor. Where the white ball went afterward was never calculated, and it quite often ended up in a pocket.

So why was it a problem? Why did they think the AI was “too good” when it was actually random?

Humans have a tendency to anthropomorphize AI opponents. We think the computer is going through a thought process just like a human would do in a similar situation.

When we see the ball end up in an advantageous position, we think the computer must have intended that to happen.

The effect is magnified here by the computer’s ability to pot a ball from any position, so for the computer, all positions are equally advantageous.

Hence, it can pot ball after ball, without having to worry about positional play. Because sinking a ball on every single shot would be impossible for a human, the player assumes that the computer is using positional play.

Design or Code?

Is this a design problem or a code problem? To a certain extent it depends on the type of game, and to what extent the AI-controlled opponents are intended to directly represent a human in the same situation as the player.

In a head-to-head game such as pool, chess, or poker, the AI decisions are very much determined at a pure code level. In a one-versus-many game, such as an FPS, there is some expectation that your opponents are generally weaker than you are.

After all, you are generally placed in a situation of being one person against countless hordes of bad guys. Other game genres, particularly racing games, pit you against a field of equal opponents. Here the expectation of realistic AI is somewhere between that of chess and the FPS examples.

The more the computer AI has to mimic the idiosyncrasies of a human player, the more the task falls to the programmer. The vast majority of the AI work in a chess game is handled by programmers. Game designers would focus more on the presentation.

In an FPS, the underlying code is generally vastly simpler than chess AI. There is path finding, some state transitions, some goals, and some basic behaviors.

The majority of the behavioral content is supplied via the game designers, generally with some form of scripting. The designers will also be responsible for coding in actions, goals, and responses that emulate the idiosyncrasies of human behavior.

Heads Up!

In some heads-up games, such as chess and pool, the computer has a huge advantage over the player. Modern chess programs such as Fritz are vastly stronger than nearly all human players.

In pool and snooker games, the computer can be programmed to never miss a shot. However, people want to play against an opponent that is well matched to their skills, and so there are generally levels of AI in the game that the player can choose from.
Advertisement

The simplest way to introduce stupidity into AI is to reduce the amount of computation that it’s allowed to perform. Chess AI generally performs billions of calculations when deciding what move to make.

The more calculations that are made (and the more time taken), then (generally) the better the computer will play. If you reduce the amount of calculations performed, the computer will be a worse player.

The problem with this approach is that it decreases the realism of the AI player. When you reduce the amount of computation, the AI will begin to make incredibly stupid mistakes — mistakes that are so stupid, no human would ever make them. The artificial nature of the game will then become apparent, which destroys the illusion of playing against a real opponent.

Remember what we are trying to accomplish: We want people to have an enjoyable experience. No matter what the game, we want the players to feel challenged so that when they win, they feel a sense of accomplishment. We want them to feel that they were playing against an opponent who was really trying to beat them.

By reducing the amount of computation, we create an AI opponent that is trying to win, but has been crippled in a way that leads to unrealistic gameplay. But does the player actually care about what is going on under the hood? What if we don’t cripple our AI, but instead let it play at full strength, but have the AI deliberately throw the game?

Throwing the Game

In sports, “throwing the game” means one side makes a series of intentional mistakes that look natural, but result in losing the game. This behavior is rightly vilified by players and fans, as the agreement is that there be a contest between two equal opponents, or at least, two opponents who are trying equally hard to win.

But in computer games, it’s impossible to have an equal match. It’s humans versus machines. One side has an advantage of being able to perform a billion calculations per second, and the other has the massively parallel human brain.

Any parity here is an illusion, and it’s that illusion that we seek to improve and maintain via the introduction of intelligent mistakes and artificial stupidity.

The computer has to throw the game in order to make it fun. When you beat the computer, it’s an illusion. The computer let you win. We just want it to let you win in a way that feels good.

AI programmers need to get used to this idea. We are manipulating the game, creating artificial stupidity, fake stupidity. But we are not predetermining the outcome of the game.

We don’t set our AI with the intent to lose the game, but rather to give the human player a reasonable chance of winning. If the human plays poorly, the AI will still win, but the player will at least feel like she came close to beating a strong opponent, and thus feel like playing one more game.

Hidden Handicapping

Computer chess expert Steven Lopez (see Resources) describes how in human versus human chess, it’s acceptable for a high-ranking player to give a much lower ranking player an advantage at the start of the game by removing some of his pieces from the board before the game begins.

When the game starts, the master player and the novice player are still playing to the height of their abilities, and yet the game is more evenly balanced. The master player does not have to play “stupid” in order to give the novice player a chance.

However, humans playing against a computer do not like to be given an advantage in this way, and prefer to play the full board against an AI opponent of approximately their skill level.

The programmers of Fritz hit upon a solution that involved the AI deliberately setting up situations that the human player could exploit (with some thought) that would allow the human to gain a positional or piece advantage. Once the human player gained the advantage, the AI would resume trying to win.

At no point here is the AI actually dumbed down. If anything, there is actually quite a bit more computation going on, and certainly more complexity.

The goal of the AI has shifted from “win the game” to “act like you are trying to win the game, but allow the human to gain a one-pawn advantage, and then try to win.” The AI needs to be more intelligent in order to appear less intelligent.

Poker AI

When I programmed the AI for Left Field’s World Series of Poker, the AI computation was basically the same for each difficultly level.

The computer would calculate the odds of winning based on the known cards, and an estimate of the opponent’s hand strength based on betting history. The odds would then be used to calculate a rate of return, which would be used to decide if they would fold, call, or raise.

There were many special case rules and exceptions, but that’s the basics. The AI players would all make the same extensive computations, running tens of thousands of simulated hands through an evaluator to calculate the rate of return.

After these calculations were performed, only then would the differentiation be performed. At that point, the best players would play their best move, and the weak AI players would make intelligent mistakes.

For weak poker AI, an intelligent mistake consists of figuring out what you should do, and then not doing it, so long as not doing it does not make you look stupid.

For example, if the human player just put in a big raise, yet you know there’s a 75 percent chance your hand is the best, then an intelligent mistake would be to fold. The odds are the AI would win, yet we are simulating a weak human player, and weak human players often fold to a large raise when they are unclear on their odds.

Conversely, weak human players often call when their chances are weak. It’s a natural thing to do and allows us to reduce the strength of the AI player, without it looking artificially stupid.

These intelligent mistakes were implemented in a probabilistic manner. The fake-stupid AI would not always fold when the human player seemed to be bluffing — it was just more likely to.

This worked very well in the highly random game of poker, because the player could never tell in any individual situation if the AI was actually making a mistake.

Since the AI was still performing its full set of millions of calculations, it never made mistakes that were inhumanly stupid, but the layer of artificial stupidity brought on by increased recklessness was enough to even the playing field and give the weak and average human players an enjoyable game.

Artificial Inaccuracy

In pool and in shooters, the computer AI is blessed with an omniscient accuracy. The shooter AI knows down to the billionth of an inch exactly where you are, and could shoot your hat off your head from five miles away. Similarly in pool, the AI knows the position of every ball and can calculate where every ball will end up before it takes a shot.

When I implemented my snooker AI, it could perfectly pot any ball off two cushions, and would almost always get a perfect break of 147 every time it played (except when it potted the white due to its lack of positional play).

It was obviously not a fun opponent to play against, so even at the highest levels, the accuracy had to be reduced, and the cushion shots had to be restricted to getting out of snookers.

Simply reducing the accuracy of the AI is not always the best way to improve gameplay. As I found with the “positional play” in snooker, random outcomes that happen to favor the computer are perceived as being intentional. If the ball ends up in a good place, or the poker AI makes a lucky call and wins on the river, it can be perceived as unfair or even cheating.

So instead of reducing the accuracy, I’d suggest, as in chess, we increase the accuracy. In order to provide an exciting and dynamic game, the AI needs to manipulate the gameplay to create situations that the player can exploit.

In pool this could mean, instead of blindly taking a shot and not caring where the cue ball ends up, the AI should deliberately fail to pot the ball and ensure that the cue ball ends up in a place where the player can make a good shot.

In a shooter, the enemy aliens should not simply randomly break from cover — they should sometimes break from cover when the player is close to them and panning toward them. They should “accidentally” throw themselves into the line of fire to make the game more interesting.

Luck of the Draw

Playing against a perfect opponent is no fun. But playing against a crippled opponent is no fun either. To create more interesting gameplay, we have to introduce the concepts of artificial stupidity and intelligent mistakes.

Intelligent mistakes seem like failings on the part of the AI, but are actually carefully calculated ways of throwing the game that make it more entertaining for the player. This does not remove the challenge, as the player still has to have a certain level of skill.

For the programmer, adding intelligent mistakes is much more complex than simply reducing the accuracy of the AI, but provides a much more rewarding experience for the player.

Resources

Liden, Lars. “Artificial Stupidity: The Art of Intentional Mistakes,” in AI Game Programming Wisdom 2, Charles River Media, 2004. http://lars.liden.cc/Publications/Downloads/2003_AIWisdom.pdf

Lopez, Steven. “Intelligent Mistakes,” Chessbase News, 2005. http://www.chessbase.com/newsdetail.asp?newsid=2579

Source:

http://www.gamasutra.com/view/feature/3947/intelligent_mistakes_how_to_.php

[Neversoft co-founder West presents a thought-provoking look at improving the believability of AI opponents in games by upping their use of "intelligent mistakes", in a piece originally written for Game Developer magazine.]

The Unstructured Information Management Architecture (UIMA) is an XML-based data retrieval architecture under development at IBM. UIMA will greatly expand and enhance the retrieval techniques underlying databases, said Alfred Spector, vice president of services and software at IBM’s Research division.

UIMA “is something that becomes part of a database, or, more likely, something that databases access,” he said. “You can sense things almost all the time. You can effect change in automated or human systems much more.”

Once incorporated into systems, UIMA could allow cars to obtain and display real-time data on traffic conditions and on average auto speeds on freeways, or it could let factories regulate their own fuel consumption and optimally schedule activities. Automated language translation and natural language processing also would become feasible.

The theory underlying UIMA is the Combination Hypothesis, which states that statistical machine learning–the sort of data-ranking intelligence behind search site Google–syntactical artificial intelligence, and other techniques can be married in the relatively near future.

“If we apply in parallel the techniques that different artificial intelligence schools have been proponents of, we will achieve a multiplicative reduction in error rates,” Spector said. “We’re beginning to apply the Combination Hypothesis, and that is going to happen a lot this year. I think you will begin to see this rolling out in technologies that people use over the next few years. It isn’t that far away.

“There is more progress in this happening than has happened, despite the fact that the Nasdaq is off its peak,” he added.

The results of current, major UIMA experiments will be disclosed to analysts around March, with public disclosures to follow, sources at IBM said.

Although it’s been alternately touted and debunked, the era of functional artificial intelligence may be dawning. For one thing, the processing power and data-storage capabilities required for thinking machines are now coming into existence.

Researchers also have refined more acutely the algorithms and concepts behind artificially intelligent software.

Additionally, the explosive growth of the Internet has created a need for machines that can function relatively autonomously. In the future, both businesses and individuals simply will own far more computers than they can manage–spitting out more data than people will be able to mentally absorb on their own. The types of data on the Net–audio, text, visual–will also continue to grow.

XML, meanwhile, provides an easy way to share and classify data, which makes it easier to apply intelligence technology into the computing environment. “The database industry will undergo more change in the next three years than it has in the last 20 due to the emergence of XML,” Spector said.

A new order
Artificial intelligence in a sense will function like a filter. Sensors will gather data from the outside world and send it to a computer, which in turn will issue the appropriate actions, alerting its human owners only when necessary.

When it comes to Web searching, humans will make a query, and computers will help them refine it so that only the relevant data, rather than 14 pages of potential Web sites, match.

IBM’s approach to artificial intelligence has been decidedly agnostic. There are roughly two basic schools of thought in artificial intelligence. Statistical learning advocates believe that the best guide for thinking machines is memory.

Based in part on the mathematical theories of 18th century clergyman Thomas Bayes, statistical theory essentially states that the future, or current events, can be identified by what occurred in the past. Google search results, for example, are laundry lists of sites other individuals examined after posing similar queries ranked in a hierarchy. Voice-recognition applications work under the same principle.

By contrast, rules-based intelligence advocates, broken down into syntactical and grammatical schools of thought, believe that machines work better when more aware of context.

A search for “Italian Pet Rock” on a statistically intelligent search engine, for example, might return sites about the 1970s novelty. A rules-based application, by contrast, might realize you mistyped the Italian poet Petrarch. A Google search on UIMA turned up the Ukrainian Institute of Modern Art as the first selection.

“The combination of grammatical, statistical, advanced statistical (and) semantics will probably be needed to do this, but you can’t do it without a common architecture,” Spector said. Thinking in humans, after all, isn’t completely understood.

“It’s not exactly clear how children learn. I’m convinced it’s statistically initially, but then at a certain point you will see…it is not just statistical,” he said. “They are reasoning. It’s remarkable.”

Tononi, professor of psychiatry at the UW-Madison School of Medicine and Public Health and an internationally known expert on consciousness, is part of a team of collaborators from top institutions who have been awarded a $4.9 million grant from the Defense Advanced Research Projects Agency (DARPA) for the first phase of DARPA’s Systems of Neuromorphic Adaptive Plastic Scalable Electronics (SyNAPSE) project.

Tononi and scientists from Columbia University and IBM will work on the “software” for the thinking computer, while nanotechnology and supercomputing experts from Cornell, Stanford and the University of California-Merced will create the “hardware.” Dharmendra Modha of IBM is the principal investigator.

“Every neuron in the brain knows that something has changed,” Tononi explains. “It tells the brain, ‘I got burned, and if you want to change, this is the time to do it.’’

Thus, a cat landing on a hot stovetop not only jumps off immediately, it learns not to do that again.

The idea is to create a computer capable of sorting through multiple streams of changing data, to look for patterns and make logical decisions.

There’s another requirement: The finished cognitive computer should be as small as a the brain of a small mammal and use as little power as a 100-watt light bulb. It’s a major challenge. But it’s what our brains do every day.

“Our brains can do it, so we have proof that it is possible,” says Tononi. “What our brains are good at is being flexible, learning from experience and adapting to different situations.”

While the project will take its inspiration from the brain’s architecture and function, Tononi says it isn’t possible or even desirable to recreate the entire structure of the brain down to the level of the individual synapse.

“A lot of the work will be to determine what kinds of neurons are crucial and which ones we can do without,” he says.

It all comes down to an understanding of what is necessary for teaching an artificial brain to reason and learn from experience.

“Value systems or reward systems are important aspects,” he said. “Learning is crucial because it needs to learn from experience just like we do.”

So a system modeled after the neurons that release neuromodulators could be important. For example, neurons in the brain stem flood the brain with a neurotransmitter during times of sudden stress, signaling the “fight-or flight” response.

“Every neuron in the brain knows that something has changed,” Tononi explains. “It tells the brain, ‘I got burned, and if you want to change, this is the time to do it.’”

Thus, a cat landing on a hot stovetop not only jumps off immediately, it learns not to do that again.

Tononi says the ideal artificial brain will need to be plastic, meaning it is capable of changing as it learns from experience. The design will likely convey information using electrical impulses modeled on the spiking neurons found in mammal brains. And advances in nanotechnology should allow a small artificial brain to contain as many artificial neurons as a small mammal brain.

It won’t be an easy task, says Tononi, a veteran of earlier efforts to create cognitive computers. Even the brains of the smallest mammals are quite impressive when you consider what tasks they perform with a relatively small volume and energy input.

“I would be happy to create a mouse brain,” Tononi says. “A mouse brain is quite remarkable. And from there, it shouldn’t be too hard to scale up to a rat brain, and then a cat or monkey brain.”

Source: http://www.sciencedaily.com/releases/2008/12/081221215537.htm
Adapted from materials provided by University of Wisconsin-Madison. Original article written by Susan Lampert Smith.