CCSI

The state of PA has granted UPenn, Temple and Pitt $10 million to develop a national center on science instruction. The center is supposed to unite the Graduate School of Education at Penn with the Institute for Research in Cognitive Science, the Spatial Intelligence and Learning Center at Temple; and the University of Pittsburgh’s Learning Research and Development Center.


Story from Marketwatch:Governor Rendell Announces $10 Million Grant to Expand Science Education

Press Release from Penn: Penn and Other Institutions Receive $10 Million Grant to Establish Center for Cognition and Science Instruction

I’ve actually been avoiding this story for a week, but a colleague of mine sent it to me, so I guess it is time to post it.

Mark Changizi at RPI (and may I say: damn, do they have a good PR department) is being reported as having ‘proven that some people can see into the future’ or having ‘discovered the key to all optical illusions’, depending on who you read:

Crystal (Eye) Ball: Visual System Equipped With ‘Future Seeing Powers’ (Science Daily.com - a copy of the original press release, which can be found: at RPI)

Key to All Optical Illusions Discovered (Yahoo News)

I Can See The Future - And Apparently, So Can You (shortnews.com)

Scientist: Humans Can See Into Future (Fox News.com)

Humans can ‘foresee the future’ (Times of India)

I’m excited to hear about evidence for a neural correlate of the specious present, but let’s not get carried away with our science journalism, please! Whenever I bring the idea of the specious present up in class, I usually cite this interview with, of all people, LeBron James:

GQ-The Fast Education of LeBron James

In the third paragraph, he says “I don’t want to sound cocky when I say this, but it’s like I see things before they happen. I kind of know where the defenders are gonna be. I kind of know where my teammates are gonna be, sometimes even before they know.” Given that all human behaviors (esp. those with clean neural correlates like this story claims) are subject to standard statistical variation, is it not possible that LeBron (and perhaps most quarterbacks and point-guards) are just statistical outliers with respect to this ability?

There’s a good article in the NYTimes on research by Lera Boroditsky and colleagues on the influence of language categories on perception. If you were there, I’m sure you’d remember her presentation at the SPP in Edmonton. I think every person in the room had their hand up for a question. I’ve never seen anything like it. Anyway - I, like many of the people there, believe her research to be solid and interesting, but worry about the conclusion drawn. I won’t bias your reading, however: When Language Can Hold the Answer

I saw this one in the Post this morning.


(Click to view full-sized image at Dilbert Archive Home>

OK, so that title is a little bit of a double-entendre, and an inaccurate at that - but it was the best I could do on short notice:

‘Science Daily’ has an article interviewing Selmer Bringsjord on his creation of an artificial reasoner on second life:
Bringing Second Life To Life: Researchers Create Character With Reasoning Abilities Of A Child

and
AI-Based Virtual Child Plays in Second Life and Child-like intelligence created in Second Life

Hurray! Scientists in Italy have achieved the greatest (to my mind) challenge in AI:
Scientists Create Coffee-Making Robot

Ray Kruzweil has predicted ‘human-level’ AI by 2029. The BBC article is here:

Machines ‘to match man by 2029′

I’m not an AI skeptic - I wouldn’t be in this specialization if I were - but I get really sick of these predictions. My father tells a (possibly) apocryphal story of sitting in Marvin Minsky’s course back in 1955 (or so). Minsky gets up, hammers his first on the podium, and announces “we’ll have artificially intelligent machines within 5 years!” Sometime in the middle 80’s, Minsky was interviewed by Parade magazine (or some other such tabloid) and he was quoted as saying “we’ll have artificially intelligent machines within 5 years!” While I understand the need for popular outreach (I’ve written for Philosophy and the Grateful Dead, for goodness sakes) for both pedagogy and funding, I wish that our ’spokespeople’ would be more careful.

I have found that one of the best ways to introduce functionalism to students is to take the Mind Project’s AI program, “Larry Learner” into the classroom and engage the students. There are two versions of the program that plays a game called Last One Loses. The trick when using it in the classroom is to start with the first version — that is hardwired with the winning algorithm. The vast majority of students will insist that it is not intelligent and after giving their own reasons why, will almost unanimously agree that Larry is not intelligent and the human author of the Larry program, Tony Kuzola, IS intelligent because he had to figure out — by trial and error — what the winning moves were and then he simply encoded it into the Larry program. But now you have the students where you want them. Because the second version of Larry, the “learning” version, does itself learn by trial and error what the winning moves are and thus has the very property that was sufficient for attributing intelligence to the human being.

Now you are ready for a great discussion about what intelligence really is. Simple knee-jerk reactions aren’t sufficient any more. Yes, even the learning version of Larry may not be intelligent, but why not? Students throw out lots of ideas and then you stop, as if surprised, and say: “Wow. That exactly what Alan Turing said and he may have been the most brilliant thinker about AI that ever lived. This leads nicely into the Turing Test and the importance of being able to give rich, flexible responses to a wide range of different situations if something is to be considered “intelligent.”

The Windows version of Larry Learner is the best to use in the classroom. But if you can’t run Windows there is also a Java-script version on our website. You can get both here: Larry Learner — An AI Program (Overview)

I have also created a webpage that takes a student through all of these issues in a way that is fairly similar to how I do it in the classroom.

Intelligence, computers, and robots

I recommend that you read it to get ideas about how to use the program in the classroom, but that you do it all yourself rather than assigning this reading to the students. If you don’t want to take the time to do it in the classroom, or don’t have the technology, you can have the students read it on their own and then you can discuss it.

For the past couple of years, I’ve been referring to that little guy in Men In Black whenever I have to debunk the homunculus theory of the mind (usually on the first day of my ‘Minds and Machines’ course). So I decided to actually bring the video with me to class:

If you want to use it, the url is http://inquiry.mcdaniel.edu/videos/MIBHumunculus.swf

While looking for something totally unrelated to AI, I came across this article on asharq alawsat - ‘the Leading Arabic Daily English edition’.
Can a Machine Issue Islamic Fatwas?
There isn’t enough detail in the article to glean anything specific about the programming of the Electronic Mufti, but it does raise some interesting issues in metaethics.