Sunday, January 20, 2008

Artificial Intelligence Suicides

Did HAL 9000 go nuts and try to kill everyone?

Are Terminator robots evil?

Thinking about such questions was important to Chris McKinstry and Push Singh.

The similar self-inflicted deaths of Chris McKinstry who created a database called Mindpixel, and Push Singh who was responsible for a database called Open Mind Common Sense has left many unanswered questions in the Artificial Intelligence (AI) community. McKinstry died in Chile and Singh died in Cambridge, Massachusetts.

David Kushner has written a remarkably detailed new essay in Wired, entitled "Two AI Pioneers. Two Bizarre Suicides. What Really Happened?" about the mystery surrounding their lives and suicides. (Thanks to Patrick Huyghe for bringing this interesting article to my attention.)

In 2006, the two AI genuises killed themselves using an unusual but almost identical method.

McKinstry had unhooked the gas line from his stove and connected it to a bag sealed around his head. He was dead at the age of 38, with January 23, 2006, declared as his death date. Four weeks after Chris McKinstry died by suicide, the police were dispatched to an apartment at 1010 Massachusetts Avenue near MIT. Inside, they found the 33-year-old Singh. He had connected a hose from a tank of helium gas to a bag taped around his head. He was dead, as of February 28, 2006.

Kushner's article notes:

Amid the grieving, there were whispers about the striking parallels between Singh's and McKinstry's lives and deaths. Some wondered whether there could have been a suicide pact or, at the very least, copycat behavior. Tim Chklovski, a collaborator with Singh on Open Mind, suggests that perhaps McKinstry's suicide had inspired Singh. "It's possible that he gave Push some bad ideas," he says. (The rumors are likely to begin again: The fact that Singh committed suicide in nearly the same way McKinstry did has not been reported or widely known until this writing.)

Details have not been forthcoming from MIT. After initial reports in the media of an "apparent suicide" by Singh, a shroud of secrecy descended. [Pioneering AI researcher Marvin] Minsky and others in the department declined to be interviewed for this article. The school has long been skittish about the topic of suicide.



The Spooky Paradigm blog has pointed out (http://spookyparadigm.blogspot.com/) that the AI suicides of McKinstry-Singh are close to the plot of an X-Files episode. I dug a little deeper in that direction, to investigate the overlaps with the method imagery of McKinstry and Singh dying with the bags over their heads (which, of course, has reflective alignments to the Heaven's Gate suicides too) and that sci-fi narrative.

Regarding the X-Files link, here is a summary of the episode in question:

X-Files, Season 5, Episode 11: "KillSwitch"
Original Air Date: 15 February 1998
Two U.S. marshals, several street-level criminals, and a computer software pioneer (Donald Gelman) are all casualties of a bizarre diner shoot-out in the middle of the night. The Lone Gunmen help Mulder and Scully track down Invisigoth, a computer hacker who knew Gelman. She informs them about an artificial intelligence program created by Gelman that has deadly intentions. The only way to stop it is to feed it the "kill switch", a CD containing a neutralizing virus.

The episode was written by William Gibson and Tom Maddox. Gibson is the author of Neuromancer (1984) and its sequels Count Zero (1986) and Mona Lisa Overdrive (1988), generally considered the definitive works of the "cyberpunk" science-fiction subgenre. I find that Gibson's novel Neuromancer earned him a Nebula Award, a Hugo Award and the Philip K. Dick award - the "Holy Trinity" of science-fiction writing. Maddox is the acclaimed author of Halo.

Tom Maddox has placed the entire script of "KillSwitch" online here.

The construction of a sentient artificial intelligence, a computer program with its own consciousness is the root of this X-Files episode.

In the plot of "KillSwitch," as it concludes, two of the human computer genuises (and at one point Fox Mulder) are found with their heads encased in virtual reality masks. At the end of the show, it is revealed that their consciousness is allegedly uploaded through the computers and may have joined the AI.

I wonder what the impact of this new knowledge about the methods of their suicides will have on their AI peers? The mirror image of their deaths appears to go beyond coincidence. Will the McKinstry-Singh suicides be the models for future AI deaths?

6 comments:

ahtzib said...

Re: my post on the matter (this is spookyparadigm).

My point wasn't so much similarities between works of fiction and the actual events, but the history of the meme from the perspective of observers who in turn create the narrative.

In other words, I think the death of these two guys is quite on topic for your work on suicide copycats. But my point is that observers group these and other suicides/murders/accidents/whoknows into a plotline which does have a strong pedigree in fiction: the disappeared or dead scientists. The Wired article, in its totality, does lead one to think about whether Singh was a copycat. But that's not why it is popping up all over, it is popping up because of the immediately thrilling notion of conspiracy, or in this case, a Frankenstein scenario (such as that portrayed by Gibson's X-Files story).

Anonymous said...

I found this article by searching for "xfiles mckinstry" because I thought, maybe somebody else had been thinking about the connections to killswitch too. Very interesting. Now wouldn't it have been crazy (but I know totally impossible) if McKinstry figured out how to upload himself to the internet--it works, then he kills himself--but before dying, he sends a message to Singh about how to upload himself too? Or more realistic, they both realize that their research is only going to lead to Skynet!!! Or like the Wired article suggested--somebody from the future came back to kill both of them to stop them from inventing Skynet or something like it. This is seriously such a great topic that an X-Files movie could be about this.

Anonymous said...

Fiction is often oracular, in ways unanticipated by it's authors. Similarly, "Artificial" intelligence has the potential to upset not only a few information warfare applecarts, but calls into question the future of the apple-selling business itself. We need not place the tragic deaths of these worthy, questing minds (Mckinstry and Singh) into the realm of conspiracy fiction, or that of conspicuously marginalized fact (...microbiologists...) to appreciate what is at stake for the Powers That Be in the matter of AI. They want a genie in a bottle - one they can contain, command and control. A web-based genie, such as that under development by both of the deceased, would be a genie, not imprisoned in a bottle, but swimming freely in an ocean. Just as the authoritarian propaganda machines of yesterday (try as they might) cannot compete with a universally accessable, interactive, multimedia information network, so any pitiful, vat-grown AI could not compete with a web-born and web-based AI, nor the sunlight and air of freedom, international co-creation and transparency that would be it's source and destiny. In the deaths of these two seminal AI figures, are we witnessing the potential of such a free and flourishing AI's "web cable being cut," as it were? It's an unsettling question - the type that makes us fret, and sigh - and seek safety in the plausibly deniable threats of the fictional variety.

Anonymous said...

what do you think, anonymous? do you think someone wanted to shut down the research?

It would seem to be impossible to shut down the research, over a long timescale - anyone can get at much more data from the Web than they can ever read in a lifetime.

I'm sure terrible things have been done in the past, to destroy people's careers and sabotage research projects. But murder - are there any known examples of this?

Anonymous said...

Murder?

Came upon Push Singh while doing a related search. My question is: Has anyone considered the probability of murder? There are a lot of scientists/ great minds that suddenly turn up dead just when they are making a breakthrough.

At least in Push Singh's case, his behavior did not indicate a suicidal tendency. Was he cut short?

Both these guys died under similar circumstances while interested in the same field.

Anonymous said...

I doubt it has anything to do with murder. Many people in the field of AI are quite depressed/troubled in one way or another. (unfortunately)