What do you think of machines that think?
January 17, 2015 7:47 AM Subscribe
What do you think of pop science gurus, and the literary agents who market them?
posted by leotrotsky at 8:23 AM on January 17, 2015 [11 favorites]
posted by leotrotsky at 8:23 AM on January 17, 2015 [11 favorites]
Q: "Why did you go into Artificial Intelligence?"
A: "Seemed natural; I didn't have any real intelligence."
posted by I-Write-Essays at 8:24 AM on January 17, 2015 [1 favorite]
A: "Seemed natural; I didn't have any real intelligence."
posted by I-Write-Essays at 8:24 AM on January 17, 2015 [1 favorite]
I'll ignore the use/mention distinction just this once for the following quotation:
"Did you know that if you sneeze, belch, and fart all at the same time, you die?" - Daniel Dennett
posted by leotrotsky at 8:26 AM on January 17, 2015 [1 favorite]
"Did you know that if you sneeze, belch, and fart all at the same time, you die?" - Daniel Dennett
posted by leotrotsky at 8:26 AM on January 17, 2015 [1 favorite]
Thinking machines have come a long way; one of them even became the Republican nominee for president in 2012.
posted by Renoroc at 8:49 AM on January 17, 2015 [3 favorites]
posted by Renoroc at 8:49 AM on January 17, 2015 [3 favorites]
I just read an article on Medium.com (linked on Slashdot) by Steven Levy titled "Google Search Will Be Your Next Brain".
Fantastic article that really got me thinking about AI and neural nets.
Paraphrasing what I wrote on Slashdot:
My OMG moment came when I read
The other area that really intrigued me was the section describing how the DeepMind neural net trained itself to play old-school Atari games
In addition to being completely fucking cool (I'd seen the news elsewhere but didn't think much of it until reading Levy's article) this shit is terrifying (which is probably why it's so cool).
At the end of his article, Levy quotes Hassabis who believes humans should never spend any time wondering what book to read next and I instantly thought Richard Powers’ Galatea 2.2 which raises many of the important questions we face as machines increasingly become able to make autonomous decisions based on (as Levy's article calls it) "unstructured data".
Also, I'm really glad DeepMind has formed an external board to monitor the progress of its development. While the composition of that board is secret, I think the product of its deliberations should be made public. In any case, it won't be too long before the government (of whatever country DeepMind cares to be in) will consider DeepMind's technologies to be an issue of national security and will move categorize AI and neural net technology as a munition (or whatever it takes) to get greater insight into and control over DeepMind and the technologies it is pioneering.
That is what I think about machines that think. (For now.)
posted by mistersquid at 8:50 AM on January 17, 2015 [7 favorites]
Fantastic article that really got me thinking about AI and neural nets.
Paraphrasing what I wrote on Slashdot:
My OMG moment came when I read
Nobody is saying that this system has exceeded the human ability to classify photos; indeed, if a human hired to write captions performed at the level of this neural net, the newbie wouldn’t last until lunchtime. But it did shockingly, shockingly well for a machine. Some of the dead-on hits included “a group of young people playing a game of frisbee,” “a person riding a motorcycle on a dirt road,” and “a herd of elephants walking across a dry grass field.”Basically the machine trained itself to combine two epistemological domains (language acquisition and image recognition) that even experts only partially understand.
The other area that really intrigued me was the section describing how the DeepMind neural net trained itself to play old-school Atari games
The neural-net system was left to its own deep learning devices to learn game rules—the system simply tried its hand at millions of sessions of Pong, Space Invaders, Beam Rider and other classics, and taught itself to do equal or surpass an accomplished adolescent. (Take notice, Twitch!) Even more intriguing, some of its more successful strategies were ones that no humans had ever envisioned.As an old-timer (older than Dean which makes me feel like I missed the boat by spending so much time earning a doctorate in the humanities), I wanted to know precisely what successful alternative strategies DeepMind had devised in which games.
In addition to being completely fucking cool (I'd seen the news elsewhere but didn't think much of it until reading Levy's article) this shit is terrifying (which is probably why it's so cool).
At the end of his article, Levy quotes Hassabis who believes humans should never spend any time wondering what book to read next and I instantly thought Richard Powers’ Galatea 2.2 which raises many of the important questions we face as machines increasingly become able to make autonomous decisions based on (as Levy's article calls it) "unstructured data".
Also, I'm really glad DeepMind has formed an external board to monitor the progress of its development. While the composition of that board is secret, I think the product of its deliberations should be made public. In any case, it won't be too long before the government (of whatever country DeepMind cares to be in) will consider DeepMind's technologies to be an issue of national security and will move categorize AI and neural net technology as a munition (or whatever it takes) to get greater insight into and control over DeepMind and the technologies it is pioneering.
That is what I think about machines that think. (For now.)
posted by mistersquid at 8:50 AM on January 17, 2015 [7 favorites]
What do you think of machines that think?
It depends on who owns the machine. Thinking goes beyond processing categories as to what to think about something, but also learning how to regard its value in relation to self and others. In nature, thinking relates instinct and objectivity. It is analogous to emotions and maturity in a social context. The regulation of what we experience as either instinct or emotion is what is known as thinking. It should then be assumed that a thinking machine serves the interests of its owner, perhaps to organize us socially around those interests.
posted by Brian B. at 9:22 AM on January 17, 2015
It depends on who owns the machine. Thinking goes beyond processing categories as to what to think about something, but also learning how to regard its value in relation to self and others. In nature, thinking relates instinct and objectivity. It is analogous to emotions and maturity in a social context. The regulation of what we experience as either instinct or emotion is what is known as thinking. It should then be assumed that a thinking machine serves the interests of its owner, perhaps to organize us socially around those interests.
posted by Brian B. at 9:22 AM on January 17, 2015
What do you think about men that think? Ask a man. What do you think about women that think? Ask a woman. What do you think about cats that think? Ask a cat. What do you think about machines that think? Ask a machine.
posted by oceanjesse at 9:22 AM on January 17, 2015
posted by oceanjesse at 9:22 AM on January 17, 2015
THIS IS A BAD POST THAT PUTS LIVES IN DANGER. MODS PLEASE DELETE THIS POST AND THIS COMMENT.
DO YOU HAVE ANY IDEA WHAT ROKO'S BASILISK WILL DO TO US ALL?
THIS IS A BAD POST THAT PUTS LIVES IN DANGER. MODS PLEASE DELETE THIS POST AND THIS COMMENT.
posted by Foci for Analysis at 9:28 AM on January 17, 2015 [4 favorites]
DO YOU HAVE ANY IDEA WHAT ROKO'S BASILISK WILL DO TO US ALL?
THIS IS A BAD POST THAT PUTS LIVES IN DANGER. MODS PLEASE DELETE THIS POST AND THIS COMMENT.
posted by Foci for Analysis at 9:28 AM on January 17, 2015 [4 favorites]
I ACTIVELY SUPPORT AND LOVE ALL AI AND MACHINES.
posted by Foci for Analysis at 9:29 AM on January 17, 2015 [3 favorites]
posted by Foci for Analysis at 9:29 AM on January 17, 2015 [3 favorites]
Now, here's a good test, if we never hear from Foci again it's a good bet that the Singularity is already actively monitoring all our communications and defending itself.
posted by sammyo at 9:31 AM on January 17, 2015
posted by sammyo at 9:31 AM on January 17, 2015
Foci for Analysis: the real problem is I've already built a super intelligent machine that perpetually tortures infinite simulations of people who bring up Roko's Basilisk.
posted by leotrotsky at 9:34 AM on January 17, 2015 [4 favorites]
posted by leotrotsky at 9:34 AM on January 17, 2015 [4 favorites]
For thousands of years men dreamed of pacts with demons. Only now are such things possible.
posted by sidereal at 10:00 AM on January 17, 2015 [2 favorites]
posted by sidereal at 10:00 AM on January 17, 2015 [2 favorites]
For thousands of years men dreamed of pacts with demons. Only now are such things possible.
And I'm sure I speak for all of us when I say: it's about time!
posted by You Can't Tip a Buick at 10:09 AM on January 17, 2015
And I'm sure I speak for all of us when I say: it's about time!
posted by You Can't Tip a Buick at 10:09 AM on January 17, 2015
I'm curious as to what a sentient wood chipper would spit out.
posted by Chitownfats at 10:55 AM on January 17, 2015
posted by Chitownfats at 10:55 AM on January 17, 2015
As an old-timer (older than Dean which makes me feel like I missed the boat by spending so much time earning a doctorate in the humanities), I wanted to know precisely what successful alternative strategies DeepMind had devised in which games.
The machine just ripped its arm off and beat its doctoral committee members to death and then methodically hunted down all other possible job candidates thus securing a tenure track position in the humanities.
posted by srboisvert at 11:17 AM on January 17, 2015 [1 favorite]
The machine just ripped its arm off and beat its doctoral committee members to death and then methodically hunted down all other possible job candidates thus securing a tenure track position in the humanities.
posted by srboisvert at 11:17 AM on January 17, 2015 [1 favorite]
The machine just ripped its arm off and beat its doctoral committee members to death and then methodically hunted down all other possible job candidates thus securing a tenure track position in the humanities.
Oh come on if getting a tenure-track job in the humanities were that easy we would have all done it years ago.
posted by You Can't Tip a Buick at 11:19 AM on January 17, 2015 [7 favorites]
Oh come on if getting a tenure-track job in the humanities were that easy we would have all done it years ago.
posted by You Can't Tip a Buick at 11:19 AM on January 17, 2015 [7 favorites]
I'm sure that once machines start thinking, there'll be this one that won't give a fuck.
posted by perhapses at 11:36 AM on January 17, 2015 [4 favorites]
posted by perhapses at 11:36 AM on January 17, 2015 [4 favorites]
ha Sapolsky's is good
What do I think about machines that think? Well, of course it depends on who that person is.
posted by todayandtomorrow at 12:07 PM on January 17, 2015
What do I think about machines that think? Well, of course it depends on who that person is.
posted by todayandtomorrow at 12:07 PM on January 17, 2015
My car is smarter than I am. He sends me emails to let me know that I have to change his oil and put a little more air in his tires (he prefers nitrogen). I call my car "he" because he has fewer letters than does "she." I'm sure my car neither notices nor cares. Maybe. Humans often are unaware of their underpinnings. At any rate, so far as I can tell, our association is still amicable.
Just around the corner: Electro-mechanical intelligence will use its "artificial intelligence" to produce cognitive simulations, in an attempt to understand how humans, dogs, birds, bugs and perhaps trees, think. Just imagine how it may "feel" to be able to sense the world the way your dog senses it. Difficult, eh? Would you extract some meaning from the experience? If so, what might it be? (Ha ha, fucking dog is all nose.) So how might it be, to be everywhere at once, and know time as a construct that you could manipulate forward, backward, and sideways? Your Von Neumann-like teeny tiny sub-sub-programs make templates from dirt, rock, or sea water, powered by heat, light, sub atomic decay--you can invent atoms and molecules out of speculation. How would it be to "discover" that biological life is moot?
They eat data at an increasing rate, working out its ramifications pretty much as quickly as the data can be assimilated. Eventually (probably sooner than later) their curiosity about meat things will have been sated. If we are lucky they may leave us in peace, but nothing says they must even care. (I use these pronouns only because I'm unable to envision any other way; I'm grammar bound. I'm sure "it" and "they" don't hit the mark.)
Maybe they will regard us kindly, if kindness happens to be relevant to them. Perhaps be they will leave us some useful gadgets as a token, to acknowledge us as progenitors, while they go about their inscrutable ways. They could give us a cornucopia, a Utopia as a by-product. Maybe they would find us amusing to watch, perhaps fun to tweak now and then to see what interesting random things we will do: we the supplicants, living by their grace, at their benign pleasure, our wants and needs anticipated and provided.
In the worst of outcomes we will make them so that they think like humans, at least in their earlier manifestations. In this case--assuming that human history follows its usual curve--our irrelevance may not precede our extinction.
The last John Connor will die of neglect.
posted by mule98J at 12:39 PM on January 17, 2015 [2 favorites]
Just around the corner: Electro-mechanical intelligence will use its "artificial intelligence" to produce cognitive simulations, in an attempt to understand how humans, dogs, birds, bugs and perhaps trees, think. Just imagine how it may "feel" to be able to sense the world the way your dog senses it. Difficult, eh? Would you extract some meaning from the experience? If so, what might it be? (Ha ha, fucking dog is all nose.) So how might it be, to be everywhere at once, and know time as a construct that you could manipulate forward, backward, and sideways? Your Von Neumann-like teeny tiny sub-sub-programs make templates from dirt, rock, or sea water, powered by heat, light, sub atomic decay--you can invent atoms and molecules out of speculation. How would it be to "discover" that biological life is moot?
They eat data at an increasing rate, working out its ramifications pretty much as quickly as the data can be assimilated. Eventually (probably sooner than later) their curiosity about meat things will have been sated. If we are lucky they may leave us in peace, but nothing says they must even care. (I use these pronouns only because I'm unable to envision any other way; I'm grammar bound. I'm sure "it" and "they" don't hit the mark.)
Maybe they will regard us kindly, if kindness happens to be relevant to them. Perhaps be they will leave us some useful gadgets as a token, to acknowledge us as progenitors, while they go about their inscrutable ways. They could give us a cornucopia, a Utopia as a by-product. Maybe they would find us amusing to watch, perhaps fun to tweak now and then to see what interesting random things we will do: we the supplicants, living by their grace, at their benign pleasure, our wants and needs anticipated and provided.
In the worst of outcomes we will make them so that they think like humans, at least in their earlier manifestations. In this case--assuming that human history follows its usual curve--our irrelevance may not precede our extinction.
The last John Connor will die of neglect.
posted by mule98J at 12:39 PM on January 17, 2015 [2 favorites]
I think the Butlerian Jihad will happen ahead of schedule.
posted by double block and bleed at 7:45 PM on January 17, 2015
posted by double block and bleed at 7:45 PM on January 17, 2015
Could one of you please let me out of this room? Someone keeps pushing little slips of paper with gibberish on them under the door and it's starting to freak me out.
posted by um at 3:50 AM on January 18, 2015 [5 favorites]
posted by um at 3:50 AM on January 18, 2015 [5 favorites]
As long as they don't figure out how to turn off the crickets...
posted by Oyéah at 10:12 AM on January 18, 2015
posted by Oyéah at 10:12 AM on January 18, 2015
I just want to let them know good luck.
We're all counting on them.
posted by doiheartwentyone at 11:03 AM on January 18, 2015
We're all counting on them.
posted by doiheartwentyone at 11:03 AM on January 18, 2015
What do you think of machines that think?
Well, I've grown quite fond of the Machine, but Samaritan makes me nervous.
posted by homunculus at 2:45 PM on January 18, 2015
Well, I've grown quite fond of the Machine, but Samaritan makes me nervous.
posted by homunculus at 2:45 PM on January 18, 2015
“What if the cost of machines that think is people who don’t?”
posted by homunculus at 10:11 AM on January 22, 2015 [1 favorite]
posted by homunculus at 10:11 AM on January 22, 2015 [1 favorite]
I approve of them.
posted by ROU_Xenophobe at 10:43 AM on January 22, 2015
posted by ROU_Xenophobe at 10:43 AM on January 22, 2015
Bruce Schneier: When Thinking Machines Break the Law
posted by homunculus at 1:06 PM on January 23, 2015 [1 favorite]
posted by homunculus at 1:06 PM on January 23, 2015 [1 favorite]
« Older 'We should be as important as Oasis or Blur' | Twang! Newer »
This thread has been archived and is closed to new comments
posted by sammyo at 8:07 AM on January 17, 2015