Artificial Intelligence and Robotics blog
Artificial Intelligence
What is Artificial Intelligence?
Feb 1st
In my interactions with people, I have often discovered that most of them have little to no understanding of what artificial intelligence is. Moreover people know little about what AI’s ultimate goal is. Defining artificial intelligence is rather hard as one can tell from the many definitions available online. In this article I will try to explain what AI is and what makes it difficult to define precisely.
Historically, the term artificial intelligence was coined by John McCarthy in 1956 during the seminal Dartmouth Conference that is widely accepted as AI’s birthplace. McCarthy maintains a website at Stanford that defines AI as,
It (AI) is the science and engineering of making intelligent machines, especially intelligent computer programs. It is related to the similar task of using computers to understand human intelligence, but AI does not have to confine itself to methods that are biologically observable.
This definition is all encompassing of what AI is. First of all AI is a science. Its scientific goal is to understand the principles that make intelligent behavior possible. Second, AI is an engineering discipline. The central engineering goal is to specify methods for the design of useful, intelligent artifacts (Poole et al.)
In other words, we are not only interested in understanding artificial intelligence systems but we also want to engineer such systems. Engineered systems include physical agents such as robots but also include software agents. McCarthy’s definition also includes the study of human intelligence via the construction of artificial intelligence systems. The word “artificial” is in fact key in the definition because AI is about constructing and studying man-made intelligence that can aid in our understanding of human intelligence. AI is not necessarily about replicating human intelligence. In fact, to this day, we do not have a good understanding of what constitutes “intelligence” let alone “human intelligence.”
The difference between artificial and real intelligence is further discussed in Poole et al,
Is artificial intelligence real intelligence? Perhaps not just as an artificial pearl is a fake pearl, not a real pearl. “Synthetic” intelligence might be a better name, since, after all, a synthetic pearl may not be a natural pearl but it is a real pearl.
So how would you identify an artificial intelligence agent if you saw one or better yet do such agents already exist? To answer these questions we must be clear about what an artificial intelligence agent is given our definition of AI. An AI agent is an agent that acts intelligently, i.e., it makes rational decisions considering its circumstances. That is, an AI agent observes its environment and chooses its actions accordingly. For example, an intelligent agent that is hungry decides to cook instead of doing laundry.
Intelligent agents are already pervasive in our lives. They work under the hood to make our lives easier. For example, online search engines utilize intelligent agents that catalogue information making it easy for us to find it. Other agents work under the hood in our cars making driving safer. Large airplanes fly smoothly and safely because of intelligent agents monitoring the system continuously correcting the flight path. AI agents monitor our email and prevent spam from reaching our Inbox while others make suggestions of what movies we might enjoy watching according to our preferences.
Finally, even though, physical agents such as Robocop still only exist in the movies, there are many intelligent robots performing complex tasks that humans can’t possibly do. For example, NASA’s twin Mars rovers are continuously exploring a distant planet while smaller robots are used here on Earth to inspect oil pipelines and explore hard to reach regions deep underwater.
The future of AI is bright and even though the field has advanced by leaps and bounds during the last 50 years there is still a lot of work left to do. It is hard to predict what artificial intelligence will achieve in the next 50 years but I hope it will come closer to meeting its goals of constructing artificial intelligence agents while helping us understand the nature of our own intelligence.
Resources:
John McCarthy: What is artificial intelligence?
D. Poole, A. Mackworth, R. Goebel (book): Computational Intelligence, a logical approach.
Prof. Peter Stone: IJCAI-2007 Computers and Thought Award
Jan 25th
Dr. Peter Stone who is assistant professor in the Department of Computer Sciences at The University of Texas Austin was the recipient of this year’s Computers and Thought Award. The award is presented during the International Joint Conference on Artificial Intelligence (IJCAI) to a person selected by a specially selected review committee. Prof. Stone received the award for his contributions to robotics and more specifically for his research in multi-agent systems in collaborative and adversarial dynamic environments.
Prof. Stone has developed a number of new machine learning and vision algorithms that enable robots to play team sports and specifically soccer. His RoboCup teams have been extremely successful over the last few years something that he attributes to the development of improved reinforcement learning algorithms.
Interestingly, the award was not presented to anyone during the last conference in 2005 because the committee did not deem anyone was good enough. This makes Stone’s selection even more important and I would like to join in congratulating him for his success.
Peter gave a very interesting talk during the conference. He claimed that artificial intelligence research should be driven by what he calls “challenge problems.” He advocates that we should work towards the ultimate goal of AI (building autonomous learning agents) using a bottom-up approach, i.e., by constructing robots capable of completing specific tasks such as playing soccer. He believes that top-down approaches miss out on discovering heuristics that would make most of AI problems solvable. One of the biggest challenges for bottom-approaches are that can be very specific to the particular task and not generalize to other problems. Early in his talk, Prof. Stone made it clear that there is a middle ground where the two approaches meet and that is his battlefield. Luckily for those who could not attend the conference, a video of the talk as well as Stone’s presentation slides are available online here.
The Computers and Thought Award is very prestigious and some very notable scientists have received it over the years. Past recipients include David Marr, Tom Mitchell, Rodney Brooks, Martha Pollack, Stuart Russell, Leslie Kaelbling and Daphne Koller among many others.
Moji Intelligent Messenger to use Artificial Intelligence for protecting children online
Jan 23rd
The Australian startup Mor(f) Dynamics Pty Ltd plans to introduce a new chatterbot or chat bot that will supervise a child’s online activities creating an online environment safe from online predators. The Moji Intelligent Messenger (IM) chatterbot takes the form of a virtual pet that uses an complex artificial intelligence engine to infer the meaning of online conversations alarming the child’s parents when the conversation is deemed threatening to the child.
“Our technology allows the pet to ‘understand’ conversations so that no matter how something is said, it can detect the other person’s intentions and determine where the discussion is heading,” said Chong, Mor(f)’s managing director.
The choice of masking such intelligent monitoring software with a virtual pet interface is based on the assumption that if the character is likeable then children will want to use Moji IM without feeling that they are not living in some ‘big brother’ environment. As such, children are given the option to design their pet using their own preferences and imagination. Moji IM is also capable of improving over time its owner’s model based on past activity.
A beta version of Moji IM is scheduled for release in May.
There is a longer article on Moji IM at The Star Online:TechCentral.
The evolution of games versus that of robots
Dec 3rd
Marshall Brain recently posted on his Robotic Nation Evidence blog a video showing the evolution of games in the last 25 years. The video is great and you should watch it if you haven’t seen it yet. Surprisingly, Marshall makes the following bold claim,
All of that evolution has occurred in about 25 years. Robots will be evolving that quickly over the next 25 years.
I would like to argue that he is in fact wrong and there is absolutely no parallel that can be drawn between the evolution of games and robots.
In my opinion, he is comparing apples and oranges. When it comes to games and especially computer games, we already know the basic mathematics for rendering surfaces. We know how to manipulate 3D structures and render polygons on a screen using projective geometry. We have had this knowledge since long before computers existed. What has happened in the last 25 years is that computers have increasingly become faster and faster allowing us to render more polygons per second than ever. I am not trying to say that there has not been lots of great research and progress in computer graphics over the years but the evolution of graphics hardware has been a major factor to the development of the real-time photorealistic scenes in computer games today.
When it comes to artificial intelligence and robotics, unfortunately we still don’t have the mathematics that we can build on. Computers have become very powerful over the years but we are still only capable of solving the smallest of AI problems. In robotics, other than faster computers, the rest of the hardware has not developed accordingly. Take for example power sources. Robots today still rely on standard type battery packs that only permit for short operational times usually no more than 1 hour. Additionally, most robots rely on electric motors for locomotion and we are not that close to developing anything as good as human muscles. Finally, there are some artificial intelligence researchers who believe that we have had for years enough computational power to build artificial intelligence agents but we lack the models to do it. Having worked on robotics and AI for almost 10 years, I can tell that having faster computers every year has done little to improve our ability to do such things as teach a robot the ability to navigate a dynamic environment.
In other words, I think that Marshall is wrong and there is absolutely no way that you can infer that robots will evolve over the next 25 years as quickly as games based on the evolution of the latter.
For what it’s worth, here is the video:
Artificial Intelligence to rule the stock market?
Nov 24th
The New York Times has published a very interesting article about how artificial intelligence programs are being used for stock-picking. In the article, well known AI researcher and futurist, Ray Kurzweil presents his view of the future when the stock market will be completely ruled by smart programs. Kurzweil believes that software can study large amounts of historic stock market data and discover patterns that a person could never hope to find. In addition, software can make trading decisions in only a fraction of the time required by human brokers thus making it possible at times to beat the market.
The truth is that smart programs have been used for many years to analyze the stock market and make predictions. Often, the results are presented to human brokers who use them to make the final decision. Kurzweil suggests that soon humans will be completely removed from the loop. One the other hand, some business people and computer scientists argue that artificial intelligence agents will not replace brokers any time soon. For one, software agents only analyze stock market data ignoring such obvious sources of information such as company press releases, social trends and human behavior. Computer scientists also realize that current state of the art data mining algorithms are not capable of handling the large amount of stock market data available. It might take years for a computer to analyze all of it.
I have to agree that I find Mr. Kurzweil’s predictions a bit too ambitious. However, I do realize that artificial intelligence agents are a big part of stock trading today and they will continue to be in the future.
You can read the complete article here and reach your own conclusions.
Marvin Minsky’s new book explores commonsense artificial intelligence
Nov 13th
Marvin Minsky is a professor at MIT, cofounder of MIT’s Artificial Intelligence Lab and one of the most well known artificial intelligence researchers alive today. Minsky is best known for his 1985 book “The Society of Mind.” In his book, Minsky, driven by results from child psychology and artificial intelligence, proposed that intelligence is the result of a large number of simple interacting agents as opposed to a monolithic and complex agent.
Bits and pieces of this theory emerged in papers through the 70s and early 80s. Papert turned his energies to applying these new ideas to transforming education while Minsky continued to work primarily on the theory. In 1985, he published “The Society of Mind,” a book in which 270 interconnected one-page ideas reflect the structure of the theory itself. Each page either proposes one such mechanism to account for some psychological phenomena or addresses a problem introduced by some proposed solution of another page.”
Recently, Minsky published a new book titled “The Emotion Machine.” In his new book, he shows how feelings, goals and emotions motivate and regulate the multiple interacting agents described in “The Society of Mind.” Minsky argues (among other things) that people solve problems by analogy and not logic. That is, we solve problems by searching for similar past experiences and then adapt our previous behavior to the new problem.
The book is available at your local bookstore but Minsky has also made available online an earlier draft. You can find it at his homepage here. This is a good book to add to your Christmas shopping list.
SRI International’s AI center is 40 years old
Oct 18th
It was October of 1966 when the AI center at SRI International was first formed. Within its first few years of existence, it was already innovating in AI and Robotics with the construction of Shakey, the first intelligent autonomous robot. Shakey was the first robot to sense and reason about its surrounding environment using the STanford Research Institute Problem Solver (STRIPS.) Shakey operated within a laboratory environment that was mostly engineered to make it easy for the robot to detect and reason about obstacles. Shakey was able to use its onboard camera to detect box-like obstacles and either avoid them or push them to a given position. Shakey’s STRIPS planner was able to construct complex plans consisting of a hierarchy of actions. Much of AI research today is still concerned with the same problems except the domains are larger and far more complex. You can watch SRI International’s promotional video of Shakey here.
Shakey was only the first of many revolutionary technologies to come out of this world famous artificial intelligence center. Other important innovations include the A* graph searching algorithm; Random Sample Consensus (RANSAC) for robust computation; the first hierarchical and non-linear planner NOAH; the second generation robot Flakey with real-time stereo vision; the Small Vision Module for real-time correlation-based stereo and the Centibots large scale distributed robotics system. Of course these are just a few of the center’s innovations in AI and robotics. You can learn more about the center’s history and current projects at the official website.
Artificial Intelligence for Computer Games
Aug 15th
If you are an avid or even casual player of computer games, you must have realized by now that the artificial intelligence of game characters is rather lucking compared to the computer graphics and game physics. Part of the reason for this is that most AI algorithms are not real-time and as a result not suitable for computer games. In addition, game developers traditionally lack a proper education and understanding of the AI techniques available to them.
A recently published book by Morgan Kaufmann Publishers wants to change this. The book is aptly named “Artificial Intelligence for Games,” and it is written by Ian Millington who holds a Ph.D. in Artificial Intelligence.
The publisher’s description of the book is, “Creating robust artificial intelligence is one of the greatest challenges for game developers. The commercial success of a game is often dependent upon the quality of the AI, yet the engineering of AI is often begun late in the development process and is frequently misunderstood. In this book, Ian Millington brings extensive professional experience to the problem of improving the quality of AI in games. A game developer since 1987, he was founder of Mindlathe Ltd., at the time the largest specialist AI company in gaming. Ian shows how to think about AI as an integral part of game play. He describes numerous examples from real games and explores the underlying ideas through detailed case studies. He goes further to introduce many techniques little used by developers today. The book’s CD-ROM contains a library of C++ source code and demonstration programs, and provides access to a website with a complete commercial source code library of AI algorithms and techniques.”
This book will make a great addition to the library of anyone considering a career in computer game development. I find though that it would also be helpful for the robotics enthusiasts who are trying to improve the AI of their home build robots.
Artificial Intelligence for Games publisher’s website.
Artificial Intelligence is 50 years old
Jul 24th
Artificial Intelligence as a research field was born in the summer of 1956 during a seminal workshop at Dartmouth College in Hanover, New Hampshire. It was just a year before that when Marvin Minsky, Nathaniel Rochester, Claude Shannon and John McCarthy proposed that they should hold a workshop to put together a roadmap about how to make machines think and learn similarly to humans. The ultimate goal was to discover computational models in order to enable machines to do commonsense reasoning. Today, John McCarthy is rightly considered the father of AI. I should note that the term “Artificial Intelligence” appeared for the first time in the proposal put forth by the previously mentioned scientists. And so this new discipline that would eventually captivate everyone’s imagination was born.
Artificial Intelligence had its ups and downs in the last 50 years. Early success solving small problems in simulation ignited a flurry of predictions about super intelligent machines taking over the world before the coming of the 21st century. Hampered by a lack of a good understanding of how commonsense reasoning works in people and a lack of computational resources, computers being very slow up until the mid nineties, AI research stalled in the 80s. Many people rushed to dismiss it as nothing more than hot air.
However, science is all about proposing and testing new theories in order to find the best ones. Since the mid-90s, AI research has advanced by leaps and bounds. We now have a better understanding of how the human brain works and that has helped us to find and test better computational models for AI. These in turn have also helped us to better understand the functions of the human brain. New techniques such as statistical analysis are helping intelligent agents to copy with large amounts of information and noisy sensors. Faster computers with vast amounts of storage are allowing us to experiment in more challenging domains and solve larger problems.
It is true that AI has not yet been able to produce a machine capable of commonsense reasoning. However, by specialization, many AI systems are actually running our world today. AI helps us fly airplanes and drive our cars. It aids doctors perform surgery. It helps us find information in the vastness of the World Wide Web. It helps us discover spam email and promptly delete it. It helps us schedule traffic lights and public transportation. It helps us analyze financial markets and make predictions about the outcome of sports events. It aids in surveillance of public spaces improving security and safety. These are only a small sample of the penetration of intelligent systems in our daily lives. Artificial Intelligence is here to stay and I bet it won’t be long before we have the understanding, methods and resources to finally construct thinking and learning machines. Let us wish and hope that such technology would only be used to benefit mankind and not destroy it.
You can find lots of information about AI’ and its 50th birthday on the Internet. However, I think that best reading about this topic is the 1955 proposal for the AI workshop. You can read it here.
Intelligent systems defeated by human incompetence
Jul 21st
It has been officially verified by two of the most important people in high tech that intelligent systems cannot cope with human incompetence.
The American Association for Artificial Intelligence held its annual conference in Boston last week. During a presentation given by web guru Tim Berners-Lee advocating the Semantic Web, Google executive and top AI scientist, Peter Norvig objected to the benefits of Tim’s web utopia,
“What I get a lot is: “Why are you against the Semantic Web?” I am not against the Semantic Web. But from Google’s point of view, there are a few things you need to overcome, incompetence being the first.”
We should note that he later clarified that by incompetent he was referring to the general user and not the presenter and his colleagues.
Anyways, it is well known that Google has been opposing the Semantic Web since the get go. They believe that the potential for abuse is too high, i.e., shady webmasters abusing the system by improperly tagging their websites in order to drive traffic to them. Google already extracts the semantics of a site by analyzing its content and not relying on the web developer to define it. Google has been fighting a losing battle against webmasters that trick Google’s algorithm in giving their site a high ranking even if its content is irrelevant to the search query; these pages are predominantly full of advertising. So far Google has been successful against these attacks by keeping their algorithm mostly secret. The Semantic Web removes this weapon from Google’s arsenal and so they are stalling at least until they can figure out how to avoid “user incompetence.”
Read more about this story from ZDNet Australia
Recent Comments