Other Side of The Singularity
The technological singularity is a point in time when a self-aware artificial intelligence (AI) recursively increases its own intelligence, leading to an “intelligence explosion” of unimaginable scale. Some people consider this project to be the fastest way, perhaps even the only way, for us to solve our most serious problems. As a species, we may not be smart enough 
to solve the big problems of the human condition, such as war, psychopathy, environmental sustainability, etc., so the idea is to create “super intelligences” that will show us how to fix the potentially civilization-destroying problems we face.
Much of the thinking and effort into strong AI development concerns the obvious risk of the project — how do we ensure that the resulting super-intelligence will be friendly? Or, as Johnathan Goldstein puts it in his interview with AI and robotics researcher Professor Noel Sharkey:
… twenty years from now, you think it’s more likely that a robot will be changing my bedpans than chasing me down the street, with lasers coming out of it’s eyes?
Dr. Sharkey thought that the latter scenario would be very unlikely.
There is another aspect of this super intelligence project that seems to get little mention, at least in the popular literature: what will it be like for them? Will the AIs be able to communicate with us, or for them, would it be like trying to explain calculus to a pigeon? How long will they try before giving up, if indeed at all?
That idea is artfully expressed in the following little piece of creative genius. It is from one of my favorite podcasts, CBC’s Wiretap, with Jonathan Goldstein. This is from the November 21st episode, The Answering Machine, where Goldstein played a reading of “Spirals”, a short story from David Eagleman:
Excerpt from Wiretap: Spirals by David Eagleman
I highly recommend that you listen to the whole episode, which includes the interview with Professor Sharkey and other humourous AI-related material. You can get it here, or in iTunes.
I’ll be attending the Humanity+ Summit next weekend, where one of the leading proponents of strong AI, Ben Goertzel, is speaking. Hopefully I’ll get a chance to ask him about how communication with a super-intelligence will be possible, given the gap.
Want on board the iPhone Gold Rush?
Click Here.
Nick on November 30th, 2009 @ 08:21
Interesting post!
I keep wondering how we could justify shutting down a self-aware entity?
What of its rights?
Are we obligated to help it survive? Indefinitely, since it would have no natural life-span?
Could we compel it to work for us without entailing slavery?
I think these questions are being overlooked in the hunger for more powerful digital tools.
We should be wary of the ethical ground we tread when creating an intelligence. My fear is that we will reach a crucial breakthrough before any real ethical guidelines are reached and before any resulting policies are enacted.
I fear the danger to ourselves, should we become happy slavers…
Brent on November 30th, 2009 @ 22:51
Good questions, Nick! They call into question the very nature of ethics: is self-awareness a sufficient condition for moral rights? Or is being alive, in the biological sense, necessary?
If you think the latter, that may be a problem in the future, when humanity merges with non-biological — or at least, questionably biological — technology. This won’t happen all be at once. One might add some artificial blood cells first, then upgrade the liver, enhance the nervous system… with each step, improving over nature. Is each step also a step toward exclusion from the ethical community? It would seem unlikely.
However, suppose we figure out how to completely virtualize our consciousness into a different substrate? The “uploading your mind to a computer” scenario. Identity questions aside, this would seem put us in the same realm as the self-conscious AI. So would your questions have more pertinence if the self-aware entity was once a human biological being?