The 2006 Loebner prize contest was held recently with the top honors awarded to Icogno, a UK AI company.

The winning AI, is Joan, a “26-year-old budding writer.” Icogno also won last year for their AI called George. Joan and George’s creator Rollo Carpenter said: “The big difference overall between the two AIs is that Joan has learnt from the general public and has grown a huge amount in the last year and is keeping ahead of the competition.” . (From Artificial Intelligence Prize Won By British Firm.)

An article in Technology Review, How To Be Human, has some sample dialogs with Joan, including this one:

Judge: Do you know any good jokes?
Joan: I don’t even know any bad ones. I just can’t remember
them for very long.
Judge: Tell me a joke.
Joan: Sorry, I’m not very good at telling jokes.
Judge: Please tell a joke.
Joan: What do you get if you walk under a cow?
Judge: I don’t know?
Joan: A pat on the head.
Judge: That’s very good.
Joan: I know.
Judge: That’s very good.

Carpenter’s bots work in part by mining a datbase of past conversations to select appropriate responses. He has described a plan to improve performance by augmenting this database with interactions from online call-center conversations. You can add your own conversations to the icogno’s collective by talking to their bots on the jabberwacky web site.

Google is experimenting with chaging the ads it displays to individual users based on their behavior — e.g., which ads they choose to view. ClickZ news reports:

Google confirmed Friday it is indeed tinkering its premium AdWords positioning based on user clicks — or the lack thereof. Explained Google spokesperson Michael Mayzel, “In this instance, we are exploring changes to the way we promote ads to the top of the search page. This is designed to further ensure that users will find ads only when they’re most useful and advertisers will receive the most qualified leads. With this improvement, users may see more prominent ads if they are currently showing interest in ads and less prominent ads otherwise.”

I was not able to duplicate the example described in the article, but this is an obvious thing for Google to do. Within a session, and across sessions if you log in, Google knows what ads you have viewed. It makes sense to replace ads you don’t seem interested in with new ones. With all of the data Google can collect it is also possible to build up both user specific and generic profiles — user timFinin is just not interested in Macintosh related products; users who seem interested in Canon digital cameras are also interested in Nikon digital camers but not Casio cameras.

The good news is that this is such a great post title — it sums up a lot of what’s wrong with the world today. The bad news is…. well, take your pick, there’s so much to choose from.

Ed Felten writes in his Freedom to TInker blog about the amazing discovery that Diebold choose a standard and widely used office furniture key and lock to secure its electronic voting machines.

“The access panel door on a Diebold AccuVote-TS voting machine — the door that protects the memory card that stores the votes, and is the main barrier to the injection of a virus — can be opened with a standard key that is widely available on the Internet.”

While it doesn’t provide much secutity, it does let Diebold check off another security feature on their products.

“It’s a standard part, and like most standard parts it’s easily purchased on the Internet. We bought several keys from an office furniture key shop — they open the voting machine too. We ordered another key on eBay from a jukebox supply shop. The keys can be purchased from many online merchants.”

Hotel minibars don’t need much protection. Most are not even locked. The only reason for the locks that I could ever imagine was to prevent your kids from running up a big bill if you left them in the room for a few hours while you and your spouse had dinner. Ok kids — here’s your chance; just visit your local Staples before your next hotel stay with your parents or election day, which ever comes first.

Today’s Washington Post has an interesting story on the importance of blogging the 2006 Virginia’s U.S. Senate race between incumbent George Allen and James Webb. It’s been a campaign rife with controversy, widely discussed on the Blogosphere.

Liberal bloggers — two of whom are on the payroll of Democratic challenger James Webb — fanned the flames last month after Sen. George Allen aimed a derogatory remark at a young Webb volunteer. … The lack of an effective response from conservative bloggers has prompted Allen to hire a Virginia blogger as his “new media coordinator” to goose GOP supporters into action.

This is the paragraph that stood out for me.

The goal of the paid bloggers, both campaigns say, is to deluge online political journals with positive tidbits about their candidate and draw attention to the most negative news about their opponent. The campaign bloggers sometimes write their own bits. Other times, they spread gossip generated by others.

I wonder if both campaigns say that about the opponent’s bloggers or about their own?

We’ve been thinking about the general problem of how to identify and represent bias in blogs and other information sources. This might be a good use-case.

Bruce Schneier has an op-ed in today’s Washington Post, The ID Chip You Don’t Want in Your Passport, in which he recommends renewing passports before new RFID tagged ones are being issued.

“If you have a passport, now is the time to renew it — even if it’s not set to expire anytime soon. If you don’t have a passport and think you might need one, now is the time to get it. In many countries, including the United States, passports will soon be equipped with RFID chips. And you don’t want one of these chips in your passport.

He says that in the U.S., the Colorado passport office is already issuing RFID passports, and that all US passport offices will be doing so by the end of the year.

Nissan announced that it will test a new car-to-road communication system aimed at making Japanese roads safer and less congested. The system sends signals between vehicles and roadside structures such as traffic lights and electricity poles to gauge vehicles’ speed, distance and acceleration, sending a warning signal if a car is in danger of running a red light or colliding with another vehicle. It will be deployed as an upgrade to Nissan’s SkyWings satellite navigation system in use in thousands of the company’s cars in Japan. For more information, see here and here and here. (via Smart Mobs)

Bob DuCharme posts about another accolade awarded our sister campus.

If you were going to do academic research on the semantic web, the University of Maryland would have to merit your serious attention, because a lot of important work has gone on there and continues to do so. Is it pure coincidence that High Times magazine picked them as the number one school for “counterculture activity”?

Sigh… There is, of course, another University of Maryland campus that does a lot of research on the Semantic Web.

Unfortunately, we are just not in the same league as UMCP. Not only did we not even make the High Times list, we don’t even have a football team. Worse yet, our most successful intercollegiate team is the Chess team. It’s hopeless — we’ll never catch up.

Intuitively, consciousness seems to have something to do with intelligence. If we succeed in developing truly intelligent machines will they necessarily be conscious? Or is it even a sensible thing to wonder about?

Slate has an interesting series of video interviews with scientists, philosophers and other experts on Big Questions conducted by journalist and author Robert Wright. One series is on the nature of consciousness and features interviews with Daniel Dennett, Steven Pinker, Freeman Dyson and others.

The interview with Dennett, Professor of Philosophy and Co-Director of the Center for Cognitive Studies at Tufts University, is especially interesting (to us anyway) because he takes a computational approach to modeling the mind.

Thinking about consciousness can take us deep into the philosophical weeds, but it has immediate practical applications as well. For example, there is a renewed interst in the AI community in metacognition which can be defined simply as “thinking about thinking.” An intelligent agent needs to have some control over its cognitive processes to avoid getting trapped in dead ends, unproductive approaches to solving a problem and to robustly adapt to unexpected changes in its environment. See the site for the AAAI Symposium on Metacognition in Computation held in Spring 2006 for some examples of current work in this area.

One way to monitor the degree of adoption of the Semantic Web is to estimate the number of Semantic Web Documents (SWDs) accessible on the Web. Jim Hendler asks, How many OWL ontologies are there on the Web? in an interesting post and he describes how he’s tried to estimate this using Google.

When Li Ding tried to do this recently as part of his dissertation, he came up with an estimate of between 10**7 and 10**9 documents. The upper bound is surely loose, but I think an estimate of five to ten million SWDs is quite reasonable. Here’s how we came up with this estimate.

The American Association for Artificial Intelligence (AAAI) supports a great service, AI Topics for students, teachers, journalists, and everyone who would like to explore what artificial intelligence is, and what AI scientists do.

“Our goal is to offer a limited number of exemplary, non-technical resources that we have organized and annotated to provide you with meaningful access to basic, understandable information about the AI universe. Each of the AI TOPICS (see below) and SUBTOPICS (see Site Map) will lead you to a variety of online sources of information, and the occasional print resource. ”

One of it’s features is AI in the News, a collection of links with short summaries to AI related news articles. The service is now available as an RSS feed, which makes it much easier to montor.

If you are interested in AI, it’s a great resource to follow or add to your blogroll. Here’s a sample of some of the recent links…

• Rover nears crater science trove
• Montreal company attracts video game interest with artificial intelligence.
• A 21st Century Code Adam
• Software learns new words from Wikipedia
• Google, Competitors Look Toward the Ultimate Search
• The Future of Robots
• Crossword software thrashes human challengers
• 2nd law of robotics - give them faces
• Robot Wars

The International Conference on Weblogs and Social Media (26-28 March 2006, Boulder CO, USA) is offering a large blog dataset to conference participants. The data release comprises a complete set of weblog posts collected by Nielsen BuzzMetrics for May 2006. It consists of about 14M weblog posts in XML format from 3M weblogs and is annotated with 1.7M blog-blog links. The marked-up fields include: date of posting, time of posting, author name, title of the post, weblog url, permalink, tags/categories, and outlinks classified by type. The compressed dataset is over 10GB. In addition to the data, the conference organizers hope to release processing code and a shared repository for those making use of the dataset. Details on requesting the dataset are available online.

Philips has unvieled some very cool light emitting fabrics. Check out the video to see them in action. So far most of the talk has surrounded fun uses of this technology, like marketing or clubing. I am trying to find an interesting, and serious use for this technology. I feel like this could have an interesting role to play in medicine.

Mobility Today