GHawkins’s IT450 Blog

Source: Wired 02.08 “Like Minds”

David Kushner follows a series of strange coincidences in the lives of Push Singh and Chris McKinstry.  Though these men had different backgrounds, they shared a passion for the advancement of artificial intelligence.  Both relied on extensive databases to “teach” their machines.  Also, “Both Canadian.  Both coders.  Both obsessed with tapping the Web to create a true artificial intelligence.  And both found dead in the same strange way.” 

McKinstry hoped ”to train a neural network into something resembling human using a database of binary propositions.”  In other words, he wanted to use questions with “yes/no” responses to “teach” his machine.  Famed AI researcher, Marvin Minsky, allegedly acknowledged McKinstry’s dream could be possible with a large enough training corpus.  This lead to  Mindpixel- McKinstry’s collaborative AI database.  He asked people to help by submitting questions and answers on-line.  Participants were awards shares in Mindpixel’s stock in exchange for their submissions.  McKinstry believed that enough (about a billion) valid submissions could “be combined to create a society of mind,” and ultimately “create a functioning digital brain.”

Singh “teamed with Stanford researcher David Stork to create a database of commonsense knowledge through open submissions.”  This project was called Open Mind Common Sense, and it also depended on on-line users to build its database.  Instead of “yes/no” questions like Mindpixel, Open Mind Common Sense “compiled factual statements like ‘every person is younger than their mother.’”  Singh believed this project would help understand the structure of a human mind.  He even suggested that giving a computer common sense could solve all the world’s problems

Singh believed Mindpixel’s interface paled in comparison to his own, and McKinstry conceded.  However, McKinstry pointed out a much more serious flaw in Open Mind Common Sense – data validation.  Mindpixelrelied on participants to verify one another’s entries; this is part of their agreement in order to receive shares.  Open Mind Common Sense, on the other hand, had no validation mechanism.

Singh and McKinstry were both making progress in the field of artificial intelligence research, unfortunately, neither got to see their project’s to the end.  First, McKinstry committed suicide.  He had a history of mental health issues, and had attempted suicide in the past.  He even posted something like a play-by-play suicide note on his website.  The weirdest part was that he tried to overdose on pills, but was found with his stove’s gas line connected to a bag around his head.  This part wasn’t so weird until four weeks later, when Singh was found dead with a hose connecting a helium tank to a bag around his head.  He did not share McKinstry’s troubled past, and the article did not make mention of a suicide letter, but it is being treated as one.  These two men were not friends – they shared a handful of emails over the years, but nothing ever developed out of them.  The coincidences are eerie – especially considering their somewhat unusual area of study. 

Source: Wired 02.08 “Unmasked by Facebook”

There is a fine line between data gathering for marketing purposes and invasion of privacy.  With the increasing popularity of on-line purchasing, one may be surprised to discover how much companies know about them.  There are many thriving companies whose service is to provide businesses with leads (potential buyers).  Scott Brown of Wired describes some of these companies as tracking regimes who “dissect you site by site, purchase by purchase.”

The term “identity harvesting” sounds uncomfortably similar to “identity theft” to me, but there is a key difference: the perpetrator.  Identity theft is an act of criminals whose main purpose is to steal money from victims.  Identity harvesting is an act of marketers whose main purpose is to persuade you to spend your money on specific products.  Both perpetrators seem to share a common goal of remaining anonymous.

Its easy to understand why criminals want to remain anonymous, but why do marketers care?  Don’t they think people would willingly submit their on-line purchase history to anyone who cared to know it?  Not me.  I must not be the only reluctant buyer, or the tracking regimes wouldn’t be surviving.  We must be agreeing to broadcast our on-line history as part of some fine print somewhere, or it could easily seen as invasion of privacy.  Then again, if there is not a line being crossed already, why is it slipped into the fine print? Facebook was called out for compiling users’ purchase histories, so they backed off… maybe we should all keep shouting.

Source: Wired 01.08 “The Last Word: Getting involved with a cheater spells t-r-o-u-b-l-e. (27 points)”

The Scrabble Word Finder is an online application used to generate high scoring scarbble words.  I imagaine this application has a database with very sophisticated search operations.  You type in the letter tiles you have and it will tell you the highest scoring move available.  This sounds simple, but the database has to consider how many tiles you have, how many spaces are available, what letters are stationary on the board, where special scoring tiles are loacted on the board, and ultimately, what combination you should use for maximum points.  Apparently, the program even gives strategic suggestions (such as saving a “u” in case you get a “q”).  In my opinion, this is what databases are about… you can store data on paper, but electronic databases allow you to access and manipulate the data in ways that are tedious or impossible without them.  This database has been taken so far that it can practically play Scrabble by itself.  Although this application changes Scrabble, an intellectual and strategic game, into a data entry exercise, it serves as an excellent example of how databases can make life easier.  With a little more artificial intellegence thrown into the mix, we will have programs telling us how to live our lives from day to day.