1263: "Reassuring"

This forum is for the individual discussion thread that goes with each new comic.

Moderators: Moderators General, Prelates, Magistrates

User avatar
Whizbang
The Best Reporter
Posts: 2238
Joined: Fri Apr 06, 2012 7:50 pm UTC
Location: New Hampshire, USA

Re: 1263: "Reassuring"

Postby Whizbang » Wed Sep 11, 2013 5:04 pm UTC

And of course there will always be the "living being" thing we can hold over their CPUs.
Image

mschmidt62
Posts: 46
Joined: Fri Mar 26, 2010 4:09 am UTC

Re: 1263: "Reassuring"

Postby mschmidt62 » Wed Sep 11, 2013 5:11 pm UTC

mathmannix wrote:
Well, Cueball didn't let the computer finish that one. I think the next word was 'composed', not the start of a new "parable". So, computers will never enjoy a salad composed of ... what? I submit it might have been saying

Computer wrote:Computers will never enjoy a salad composed only of bland iceberg lettuce and a dollop of 'French' dressing.


Because really, why would a computer enjoy that? It takes a human.


I read it as "Computers will never enjoy a salad composter." Because why should a computer care about recycling nutrients for biological systems?

commodorejohn
Posts: 1182
Joined: Thu Dec 10, 2009 6:21 pm UTC
Location: Placerville, CA
Contact:

Re: 1263: "Reassuring"

Postby commodorejohn » Wed Sep 11, 2013 5:28 pm UTC

At least humans are still better at freaking out over the idea that they could ever be rendered obsolete by a form of existence that requires an advanced industrial manufacturing infrastructure simply to come into being.

Image

(That's right, computers! We can make more humans using only humans and food!)
"'Legacy code' often differs from its suggested alternative by actually working and scaling."
- Bjarne Stroustrup
www.commodorejohn.com - in case you were wondering, which you probably weren't.

KarenRei
Posts: 285
Joined: Sat Jun 16, 2012 10:48 pm UTC

Re: 1263: "Reassuring"

Postby KarenRei » Wed Sep 11, 2013 5:29 pm UTC

Another limiting factor is our ability to write software. It seems to me that computers could be doing a lot more than they are doing today, if only we had the software. There isn't a program without bugs, lots of things are possible but no one ever implemented them, and many of the cool things that we can do only exist as isolated applications, instead of being linked together.


Indeed. In fact, it sounds weird but even *programming languages* are largely to blame and are evolving to work past it. In the olden days, writing anything with parallelism took a lot of effort even if you were just doing an extremely simple task. Nowadays we're getting features built into programming languages. In c++11 you can std::thread and detach a lambda function in a single line. Oh, and let's not forget about std::async and futures, where you can run things in their own thread, then use their return values later, and you don't have to worry about whether it's returned yet, as waits get inserted automatically for you where needed. The threading functions (at least in theory) even figure out on their own what's best, to actually run something in a new thread or just inline it, so you don't need to worry about trying to thread inefficiently or making too many threads. And of course we have all of these threadsafe operations built standard for std types.

The fact that threading was a pain before meant people didn't use it as much as they should have, which meant that hardware had to focus on getting as much speed out of each core as possible instead of massively parallelizing cores, which greatly limited potential performance. The most parallell power most computers have today is the GPU, but what percent of programs do you think dispatch processing work to the GPU? It's usually just too much effort.

Basically, assuming programmers keep doing more and more with threading, and hardware developers follow suit, we open up dramatically more processing capability. Neurons aren't particularly fast. They're just *massively* parallel. A modern processor core can do billions of operations per second. In practice, in typical usage, what percent of those billions per second you think must be executed in *exactly* that order? I mean of course we can all think of *specific* tasks that can only be done with in-order operation on every last element, but most of the time when you're doing so much work, you're processing large amounts of data, and at least some of the steps on that data can be done in parallel - thousands, millions, even billions of parallelisms. And the programming goal is that the programmer can order it parallelized without having to worry about the nitty-gritty implementation details.

Thousands of cores on a GPU? That's cute, but I look forward to *millions* of cores to compete with the hundred billion neurons in the brain. And it'll happen one day.
Last edited by KarenRei on Wed Sep 11, 2013 5:40 pm UTC, edited 1 time in total.

slesfo
Posts: 2
Joined: Sat Oct 03, 2009 6:00 am UTC

Re: 1263: "Reassuring"

Postby slesfo » Wed Sep 11, 2013 5:39 pm UTC

Nobody has mentioned Arimaa yet? The game designed so that humans are better than the computers?

User avatar
Pfhorrest
Posts: 5448
Joined: Fri Oct 30, 2009 6:11 am UTC
Contact:

Re: 1263: "Reassuring"

Postby Pfhorrest » Wed Sep 11, 2013 5:55 pm UTC

Soteria wrote:despite the lack of evidence for computers to ever be able to do something they weren't programmed to do, people keep on believing in true AI.


Nobody expects computers to do things they weren't programmed to do.

We expect computers to eventually be programmed to do everything a human can do.

I agree with other commentators that this is not a problem of computing power however, this is a problem of software design. We don't even fully understand what exactly it is that humans do on the higher levels that are interesting to us. We're learning more and more about how the hardware of the human brain works but that's like trying to understand some piece of complex software (say something that can recognize voices, translate them into another language, and synthesize new speech from them) by studying the transistors and logic gates of the computer it's running on. Except it's worse, because I can give a nice simple one-sentence overview of what that software does. Nobody seems to be able to agree on what exactly it is that humans do that's unique and special to conscious, thinking things; what exactly do we want a computer to emulate in the first place?

Until we understand what it is exactly that humans do which makes them uniquely human, until we can precisely describe whatever je ne se qua that I have but my laptop does not have, we can't hope to program a computer to do that. We're not only unable to figure out how to [make a computer] do something, we're unsure of what exactly it is we're trying to [make a computer] do in the first place.

But even if we eventually figure out what exactly it is we require of them to count as conscious, truly thinking machines, they will still be that way because we built and programmed them that way. Which isn't anything against them, it's just... nobody expects otherwise, so what's your point?
Forrest Cameranesi, Geek of All Trades
"I am Sam. Sam I am. I do not like trolls, flames, or spam."
The Codex Quaerendae (my philosophy) - The Chronicles of Quelouva (my fiction)

speising
Posts: 2354
Joined: Mon Sep 03, 2012 4:54 pm UTC
Location: wien

Re: 1263: "Reassuring"

Postby speising » Wed Sep 11, 2013 6:09 pm UTC

if it were a hardware power problem, we'd simply have an ai that runs very slow. we'd be very happy to accomplish that, even if it weren't practically applicable yet.

cjcampbell
Posts: 2
Joined: Wed Sep 11, 2013 6:32 pm UTC

Re: 1263: "Reassuring"

Postby cjcampbell » Wed Sep 11, 2013 6:41 pm UTC

I think it will be awhile before computers surpass humans. Somebody once asked Cecil Adams how many calories the brain uses. The fact is, not that many.

Sure, in terms of sheer processing power, machines will undoubtedly overtake humans, some say within the decade. But let’s put that in perspective. Remember, the adult brain uses about 20 watts, meaning its productivity is about 50 petaflops per watt. The output of a typical supercomputer is just under 2.5 gigaflops per watt. That’s 1/20 millionth the efficiency of the brain. To put that in terms a little easier to grasp, the Titan supercomputer is a liquid-cooled 8.2-megawatt monster filling a building the size of a large suburban house. The more powerful human model can be run on Cheerios and fits under a hat.


A good point. When we can build a computer that can beat a human at Go while using only 20 watts of power and fit the entire thing into half of a hollowed out bowling ball, all while doing all of the other things a human brain does, then I will say that computers have surpassed humans. As Darth Vader said, "Don't be too proud of your technological marvel."

Wumbo
Posts: 5
Joined: Thu Aug 23, 2012 12:10 pm UTC

Re: 1263: "Reassuring"

Postby Wumbo » Wed Sep 11, 2013 7:11 pm UTC

It's impressive that we have designed a computer which can beat the best humans in Chess and Go, but it will be more impressive when we can create computers that can win in games with significant amounts of incomplete information or that can design programs themselves. That I think is a long way off, but not impossible. There was an old xkcd comic that tried to rank the difficulty of computers for beating humans at various games, and I think this is a sequel to it. It had some games of incomplete information below Go, which I am inclined to disagree with.

Translation is a cool problem that if programmers solved it would put a lot of people out of work. Part of the problem with that is that there are varying philosophies on what translation should be, but even then we haven't be able to design a single program capable of adequately translating, despite an incredible amount of interest in the topic. The thing humans do better than computers, it's my guess, is that humans have the ability to say that something doesn't quite sound right. Our brain matches it up against phrases we've heard a million times in our life and says 'this isn't a structure I'm familiar with.' We should be able to do that with computers too, just match it up against a library of phrases, but then that only tells us that a translation is wrong. Fixing it requires understanding the elements of both languages and translating the meaning, not the words. That's what is hard for a computer to do, especially since a lot of languages have many different meanings for each word, and right now only a human can say which one is logically the correct one, because we're checking that against our memory of not just language, but of human experience. Does this exist in the real world? Have the sentences before this one been talking about this already, even if it is out of this world? Following the logic of the previous passages, what is this supposed to mean? Getting a computer to understand meaning is something we're far off from, and if anything is reassuring about that, it's that the human brain has a massive, incalculable amount of information stored in it that is hard to even translate into words, let alone code. When you read something, you're not just getting the dictionary definitions of the words, but the memories those words conjure up. You read the word cat, and you might picture the cat. You might think of your cat and that you love him. Or you might think of a cat you disliked, and that's a feeling, that's a chemical reaction in your brain that makes you feel happy or bad. It's hard to program a computer to understand that about one thing, let alone all the things we humans care about. Cat ['happy-feeling: 0.89' , 'image-location: mycat.jpg', 'memories-of: 81093' , 'cats-normally: sleep' , cats-dont: fly' ,etcetcetc] A computer needs to have all that info if it wants to translate the way a human translates. You can hardcode translation like google tries to for common phrases "the cat is cute : 猫はかわいいです。” but that's hardly going to get you far, especially when you have to take liberties with honorifics that exist in one language and not another.

If I imagine the computer equivalent to my blurb above... Imagine computers kept a log of every 1 and 0 that every passed through their precious computer brain, and that there were other computers around all keeping track of each other so that when one was shut down or crashed, it would be informed the moment it rebooted and have some idea of what had happened and what the humans had done to it. Guess how friggin massive that log file would be lool. That's what a human does. If we ever have enough space, we might be able to make computers care about things that happen in a computers life, but translating the data of what happens in a humans life would still be far off. Of course we humans forget some things, group others together, and fabricate still others entirely... And you could make the computer do that too, if you added pleasure and unhappiness into their programing (get more pleasure, avoid unhappiness... delete bad memories hehe)... But wanting a computer to remember everything it ever did is scary.

Don't mind me, I was in a rambling mood.
Last edited by Wumbo on Wed Sep 11, 2013 7:57 pm UTC, edited 2 times in total.

jpablo
Posts: 5
Joined: Wed Mar 18, 2009 6:30 pm UTC

Re: 1263: "Reassuring"

Postby jpablo » Wed Sep 11, 2013 7:31 pm UTC

We're still better at recognizing images and shapes right?

wolfticket
Posts: 40
Joined: Mon Sep 08, 2008 6:32 pm UTC

Re: 1263: "Reassuring"

Postby wolfticket » Wed Sep 11, 2013 7:50 pm UTC

We're still better at creating things that are better than us at doing other things.

User avatar
neoliminal
Posts: 626
Joined: Wed Feb 18, 2009 6:39 pm UTC

Re: 1263: "Reassuring"

Postby neoliminal » Wed Sep 11, 2013 7:58 pm UTC

Hidden information, like in poker or stratego, are where humans have a huge advantage. Games where there is total access to all information will likely get an edge to the computer once the programmers are given time to understand the problem.

Guessing is hard.
http://www.amazon.com/dp/B0073YYXRC
Read My Book. Cost less than coffee. Will probably keep you awake longer.
[hint, scary!]

Soteria
Posts: 66
Joined: Sun Jul 15, 2012 6:18 pm UTC

Re: 1263: "Reassuring"

Postby Soteria » Wed Sep 11, 2013 8:00 pm UTC

Pfhorrest wrote:
Soteria wrote:despite the lack of evidence for computers to ever be able to do something they weren't programmed to do, people keep on believing in true AI.


Nobody expects computers to do things they weren't programmed to do.

...

But even if we eventually figure out what exactly it is we require of them to count as conscious, truly thinking machines, they will still be that way because we built and programmed them that way. Which isn't anything against them, it's just... nobody expects otherwise, so what's your point?


You keep on saying nobody expects otherwise. Yet I don't think you have to look very hard on the internet to find people who think that computers will soon be able to do everything we can do and eventually take over the world.

User avatar
gmalivuk
GNU Terry Pratchett
Posts: 26767
Joined: Wed Feb 28, 2007 6:02 pm UTC
Location: Here and There
Contact:

Re: 1263: "Reassuring"

Postby gmalivuk » Wed Sep 11, 2013 8:04 pm UTC

Yes, well, you also don't have to look very hard on the internet to find timecube.com. That doesn't mean we should give it serious consideration here, nor does it contradict the fact that nobody who actually has anything whatsoever to do with this current discussion thread expects otherwise.

Oh, and we already have computers that do specific tasks they weren't programmed to do, because their programming is at a higher level which allows the computer to essentially figure out for itself what to do in a particular situation.
Unless stated otherwise, I do not care whether a statement, by itself, constitutes a persuasive political argument. I care whether it's true.
---
If this post has math that doesn't work for you, use TeX the World for Firefox or Chrome

(he/him/his)

User avatar
LaserGuy
Posts: 4582
Joined: Thu Jan 15, 2009 5:33 pm UTC

Re: 1263: "Reassuring"

Postby LaserGuy » Wed Sep 11, 2013 8:13 pm UTC

rmsgrey wrote:Another technology that's been "twenty years away" for rather more than 20 years is fusion reactors - about 5 years ago, I saw an interview with a researcher where he refined the prediction: "they're about 20 years away if they get some serious funding" - part of the problem is that people with money to apportion are interested in things that take less than 5 years to pay off, may be persuaded about things that take 10, but very few are willing to sink money into something that won't pay off for at least 20 years (apart from anything else, they tend not to expect to be around to enjoy the benefits...)


They've already started construction on a 500MW fusion reactor in France. Not saying that they'll finish it, but, at the moment, there is a funded project underway and people are actually building it. First light is expected for 2020.

[edit]
neoliminal wrote:Hidden information, like in poker or stratego, are where humans have a huge advantage. Games where there is total access to all information will likely get an edge to the computer once the programmers are given time to understand the problem.


A lot of poker theory is based heavily on mathematics and statistics though, which computers are quite good at. The University of Alberta has held a few bot vs. human poker tournaments and the results have actually been very close (humans won the first tournament and the bots won the second). I think poker is probably a lot closer to being "solved", as much as it can be, than you might think.
Last edited by LaserGuy on Wed Sep 11, 2013 8:20 pm UTC, edited 1 time in total.

User avatar
ucim
Posts: 6860
Joined: Fri Sep 28, 2012 3:23 pm UTC
Location: The One True Thread

Re: 1263: "Reassuring"

Postby ucim » Wed Sep 11, 2013 8:20 pm UTC

mward wrote:(2) "Moore's Law", observed in 1965 that computer power doubles every two years. This "law" has continued to hold for the subsequent four decades, yet despite this huge technological gains over the last 60 years or more, human intelligence is still just as far away as it ever was. It is as if despite building bigger and bigger ladders, we are getting no closer to Andromeda galaxy!
No, it's not "just as far away" as it ever was. It is simply "still too far away for us to see and recognize". Computers have become much more powerful tools than they ever were. Clear progress is definitely being made. However, our own definition of what "intellegence" is is still too undeveloped and egocentric to be useful here.

I predict that while we are trying to make a computer "be" intelligent, we will as a matter of course be using more and more powerful computer networks. When we discover how computers have "become" intelligent, it will be a surprise from a completely unexpected direction.

I have no doubt that it will happen, but I am certain that it will not be the "singularity" that is predicted. That is egocentric silliness.

I am not certain it will be a good thing, at least not for us. We'll be the frog in hot water.

I have no idea when. But when it does, we'll realize it happened some time ago, and there's no turning back.

Diadem wrote:It seems to me that computers could be doing a lot more than they are doing today, if only we had the software. There isn't a program without bugs...
Biology has bugs too. (And I don't mean that in the pun way). Organisms don't have to be defect-free. They just have to be good enough to last long enough to reproduce.

Soteria wrote:I agree, but despite the lack of evidence for computers to ever be able to do something they weren't programmed to do, people keep on believing in true AI.
However, we are programming them more indirectly now, and programming them to take more into account. It's harder to predict on a fine scale what a computer will do. They don't even boot up the same way every time, for chirp's sake!

And really, what is the difference between "intelligence" and "the illusion of intelligence", if the illusion can be kept up long enough and under varied enough circumstances?

Jose
Order of the Sillies, Honoris Causam - bestowed by charlie_grumbles on NP 859 * OTTscar winner: Wordsmith - bestowed by yappobiscuts and the OTT on NP 1832 * Ecclesiastical Calendar of the Order of the Holy Contradiction * Heartfelt thanks from addams and from me - you really made a difference.

User avatar
neoliminal
Posts: 626
Joined: Wed Feb 18, 2009 6:39 pm UTC

Re: 1263: "Reassuring"

Postby neoliminal » Wed Sep 11, 2013 8:38 pm UTC

Poker may be closer to competition but it shows the limitations of computers. Good Rock Paper Scissor players can beat computers although average players lose. This has to do with the problems of self analysis of pattern recognition in humans.

What is interesting is the programming that goes into predicting what hand the opponent has. Bluffing in poker is much more like Rock Paper Scissors and the results can be very similar in no-limit games. Computers likely have the advantage in limit poker.
http://www.amazon.com/dp/B0073YYXRC
Read My Book. Cost less than coffee. Will probably keep you awake longer.
[hint, scary!]

User avatar
Diadem
Posts: 5654
Joined: Wed Jun 11, 2008 11:03 am UTC
Location: The Netherlands

Re: 1263: "Reassuring"

Postby Diadem » Wed Sep 11, 2013 9:07 pm UTC

neoliminal wrote:Good Rock Paper Scissor players can beat computers although average players lose.

lol. You do realize right that there is no such thing as a good rock paper scissors player? Or that writing an algorithm that at least draws against any other player is trivial
It's one of those irregular verbs, isn't it? I have an independent mind, you are an eccentric, he is round the twist
- Bernard Woolley in Yes, Prime Minister

User avatar
gmalivuk
GNU Terry Pratchett
Posts: 26767
Joined: Wed Feb 28, 2007 6:02 pm UTC
Location: Here and There
Contact:

Re: 1263: "Reassuring"

Postby gmalivuk » Wed Sep 11, 2013 9:09 pm UTC

It's actually possible to make a robot that consistently beats human RPS players, which while technically "cheating" does not do so in a way that is perceptible to the human opponent.
Unless stated otherwise, I do not care whether a statement, by itself, constitutes a persuasive political argument. I care whether it's true.
---
If this post has math that doesn't work for you, use TeX the World for Firefox or Chrome

(he/him/his)

project2051
Posts: 178
Joined: Fri Jan 29, 2010 2:20 pm UTC

Re: 1263: "Reassuring"

Postby project2051 » Wed Sep 11, 2013 9:21 pm UTC

I once saw a show that compared human and computer for crossing a busy street, and how a human could without even thinking about it judge the distance across the street, track and judge speed and distance of multiple cars and pick a time and speed to cross safely. And that the computer would not be able to do so in a timely enough manner to be able to get across. I wonder if that still holds.

As for games, I wondered about Clue. That has elements of guessing, but a lot of it is basically data acquisition and collating.

User avatar
Klear
Posts: 1965
Joined: Sun Jun 13, 2010 8:43 am UTC
Location: Prague

Re: 1263: "Reassuring"

Postby Klear » Wed Sep 11, 2013 9:43 pm UTC

Diadem wrote:
neoliminal wrote:Good Rock Paper Scissor players can beat computers although average players lose.

lol. You do realize right that there is no such thing as a good rock paper scissors player?


Of course there is. Doesn't work against computers, but there are good players.

Mikeski
Posts: 1104
Joined: Sun Jan 13, 2008 7:24 am UTC
Location: Minnesota, USA

Re: 1263: "Reassuring"

Postby Mikeski » Wed Sep 11, 2013 11:32 pm UTC

Diadem wrote:We have computers that can understand spoken language, and synthesize speech. Not nearly well enough for our liking, but probably better than a 2-year-old child.

If they "understood" spoken language--if they could connect words to concepts, pick out hyperbole and metaphor, etc.--we would have reasonable translations between languages. Most machine translations I've seen sound like a 2-day-old child with the vocabulary of a novelist. :mrgreen:

Siri and Google and such are really just Chinese Rooms. Very complex ones, but not "understanding" at all.

User avatar
neoliminal
Posts: 626
Joined: Wed Feb 18, 2009 6:39 pm UTC

Re: 1263: "Reassuring"

Postby neoliminal » Thu Sep 12, 2013 12:08 am UTC

Diadem wrote:
neoliminal wrote:Good Rock Paper Scissor players can beat computers although average players lose.

lol. You do realize right that there is no such thing as a good rock paper scissors player? Or that writing an algorithm that at least draws against any other player is trivial


For a single game, perhaps. In single elimination KO tournaments you need skill.

http://www.nytimes.com/interactive/scie ... ssors.html

Play that and tell me how you do. I can beat it on both settings... I don't think you will do so well. :P
http://www.amazon.com/dp/B0073YYXRC
Read My Book. Cost less than coffee. Will probably keep you awake longer.
[hint, scary!]

Nnelg
Posts: 39
Joined: Mon May 30, 2011 4:44 am UTC

Re: 1263: "Reassuring"

Postby Nnelg » Thu Sep 12, 2013 12:30 am UTC

Whizbang wrote:And of course there will always be the "living being" thing we can hold over their CPUs.

Are you so sure? All the most important factors of life are possible for machines: self-maintenance(Homeostasis/Metabolism), response to stimuli/adaptation to their environment, growth and reproduction.

So how is any definition of "life" that automatically excludes digital sentience anything but pure racism?
keithl wrote:As a rule of thumb, it is imprudent to pass over speed bumps faster than orbital velocity.

Ronfar
Posts: 132
Joined: Mon Dec 04, 2006 6:04 am UTC

Re: 1263: "Reassuring"

Postby Ronfar » Thu Sep 12, 2013 2:34 am UTC

Here's something that you wouldn't expect humans to beat computers at, but they do: predicting protein structures.
- Doug

Turing Machine
Posts: 98
Joined: Fri Feb 19, 2010 5:48 am UTC

Re: 1263: "Reassuring"

Postby Turing Machine » Thu Sep 12, 2013 3:47 am UTC

Psst: computers still can't play chess.

User avatar
Pfhorrest
Posts: 5448
Joined: Fri Oct 30, 2009 6:11 am UTC
Contact:

Re: 1263: "Reassuring"

Postby Pfhorrest » Thu Sep 12, 2013 3:55 am UTC

Oh not this again...
Forrest Cameranesi, Geek of All Trades
"I am Sam. Sam I am. I do not like trolls, flames, or spam."
The Codex Quaerendae (my philosophy) - The Chronicles of Quelouva (my fiction)

User avatar
gmalivuk
GNU Terry Pratchett
Posts: 26767
Joined: Wed Feb 28, 2007 6:02 pm UTC
Location: Here and There
Contact:

Re: 1263: "Reassuring"

Postby gmalivuk » Thu Sep 12, 2013 4:24 am UTC

Turing Machine wrote:Psst: computers still can't play chess.

Unless, of course, you do something absurd like use a definition of "play chess" that is even remotely in line with the way every other English speaker understands those words.
Unless stated otherwise, I do not care whether a statement, by itself, constitutes a persuasive political argument. I care whether it's true.
---
If this post has math that doesn't work for you, use TeX the World for Firefox or Chrome

(he/him/his)

Air Hadoken
Posts: 16
Joined: Wed Mar 18, 2009 8:25 pm UTC
Location: San Francisco, CA
Contact:

Re: 1263: "Reassuring"

Postby Air Hadoken » Thu Sep 12, 2013 7:11 am UTC

I'm surprised that no one else actually went and implemented the script in the last panel. Inspired by today's comic, I hacked Darius Kazemi's Metaphor-a-Minute Twitter bot to make it spit out non-parable non-metaphor assertions every two minutes, and set up an account for it.

Now we shall see if I have enough historical posts to allow me to add links (to call me an "infrequent" poster is a gross understatement):

http://twitter.com/reassuring_meta

[EDIT] Yay, it worked!

User avatar
BlitzGirl
Posts: 9099
Joined: Mon Sep 20, 2010 11:48 am UTC
Location: Out of the basement for Yip 6! Schizoblitz: 115/2672 NP
Contact:

Re: 1263: "Reassuring"

Postby BlitzGirl » Thu Sep 12, 2013 8:02 am UTC

That is brilliant, Air Hadoken! :D Mostly gibberish, but there are gems in there!
Some of my favorites so far:

Computers will never mutate a planet as well as humans.
Computers will never gift-wrap a coupling as well as humans.
Computers will never distrust a superman as well as humans.
Computers will never quarrel an enigma as well as humans.
Computers will never experience a hymen as well as humans.
Computers will never discredit an angel as well as humans.
Computers will never misidentify a camel as well as humans.
Knight Temporal of the One True Comic
BlitzGirl the Pink, Mopey Molpy Mome
Spoiler:
Image
Image
Image<Profile
~.Image~.FAQ->Image

User avatar
orthogon
Posts: 3080
Joined: Thu May 17, 2012 7:52 am UTC
Location: The Airy 1830 ellipsoid

Re: 1263: "Reassuring"

Postby orthogon » Thu Sep 12, 2013 8:18 am UTC

BlitzGirl wrote:That is brilliant, Air Hadoken! :D Mostly gibberish, but there are gems in there!
Some of my favorites so far:

Computers will never mutate a planet as well as humans.
Computers will never gift-wrap a coupling as well as humans.
Computers will never distrust a superman as well as humans.
Computers will never quarrel an enigma as well as humans.
Computers will never experience a hymen as well as humans.
Computers will never discredit an angel as well as humans.
Computers will never misidentify a camel as well as humans.

"distrust a superman" and "experience a hymen" get better the more you think about them.

A lot of the gibberish is because there are nouns in the verb slot - can you (Air Hadoken) twerk it somehow? It apparently knows about parts of speech, and I realise that verbing nouns is all the rage; but coffin, servant, fellowship, maker...?
xtifr wrote:... and orthogon merely sounds undecided.

User avatar
Diadem
Posts: 5654
Joined: Wed Jun 11, 2008 11:03 am UTC
Location: The Netherlands

Re: 1263: "Reassuring"

Postby Diadem » Thu Sep 12, 2013 9:44 am UTC

Klear wrote:
Diadem wrote:
neoliminal wrote:Good Rock Paper Scissor players can beat computers although average players lose.

lol. You do realize right that there is no such thing as a good rock paper scissors player?

Of course there is. Doesn't work against computers, but there are good players.

No. Like all games of almost pure chance, you can be bad at all, but you can't be good at it. It's possible to do stupid things, and thus be bad, but once you have a minimal understanding of what to do, you will play exactly as well as everybody else.
It's one of those irregular verbs, isn't it? I have an independent mind, you are an eccentric, he is round the twist
- Bernard Woolley in Yes, Prime Minister

KarenRei
Posts: 285
Joined: Sat Jun 16, 2012 10:48 pm UTC

Re: 1263: "Reassuring"

Postby KarenRei » Thu Sep 12, 2013 9:47 am UTC

LaserGuy wrote:They've already started construction on a 500MW fusion reactor in France. Not saying that they'll finish it, but, at the moment, there is a funded project underway and people are actually building it. First light is expected for 2020.


They'll finish it - that's not the problem. The problem is that it's far from being an effective power generation method. Even the followup, DEMO, isn't expected to be commercializable. The hope is that what comes after THAT might be.

Much more short-term economical fusion power is likely to come from HiPER, if it gets funded:

http://en.wikipedia.org/wiki/HiPER

Think "NIF", but instead of relying solely on laser compression, it uses a laser heating pulse as well, so that the compression pulse doesn't need to be anywhere close to as intense. Sort of a best-of-all-worlds sort of thing. Instead of many megajoules needed to produce a laser pulse as in NIF, HiPER would need only 270kJ (the amount of energy in a laptop battery). Yet HiPER has a theoretical yield of 25-30MJ per firing (versus NIF's max of 45MJ) (both roughly in the ballpark of the energy of a gallon of gasoline). And it should be far smaller and cheaper to build than both ITER and NIF, and has a lot more potential for improvement than ITER without having to get even bigger (actually, just the opposite, things like replacing the gas lasers with diodes, as is the goal, would reduce power consumption and heat generation tenfold and allow it to become even smaller).

Like ITER and NIF, HiPER wouldn't in itself be a power plant. But it seems a much more likely route to an economical fusion power plant than either. And it's still "mainstream" fusion, not some powered-by-the-drool-of-oggling-geeks thing like Polywell or Focus Fusion.

BTW the criticism that "fusion is always 20-30 years away" isn't so much a criticism of fusion as it is with people's inability to think beyond a 20-30 year timeframe. Our fusion gain factors today are orders of magnitude better than they were 20-30 years ago.
Last edited by KarenRei on Thu Sep 12, 2013 12:09 pm UTC, edited 4 times in total.

User avatar
Klear
Posts: 1965
Joined: Sun Jun 13, 2010 8:43 am UTC
Location: Prague

Re: 1263: "Reassuring"

Postby Klear » Thu Sep 12, 2013 10:26 am UTC

Diadem wrote:
Klear wrote:
Diadem wrote:
neoliminal wrote:Good Rock Paper Scissor players can beat computers although average players lose.

lol. You do realize right that there is no such thing as a good rock paper scissors player?

Of course there is. Doesn't work against computers, but there are good players.

No. Like all games of almost pure chance, you can be bad at all, but you can't be good at it. It's possible to do stupid things, and thus be bad, but once you have a minimal understanding of what to do, you will play exactly as well as everybody else.


You underestimate how hard it is to be truly random. Unless you manage to do that, it can be exploited and turned against you.

Edit:
Read this. Furthermore, there are championships where A) some people win a lot more often than others (citation needed) and B) you wouldn't stand a chance.

rmsgrey
Posts: 3633
Joined: Wed Nov 16, 2011 6:35 pm UTC

Re: 1263: "Reassuring"

Postby rmsgrey » Thu Sep 12, 2013 10:45 am UTC

Diadem wrote:
Klear wrote:
Diadem wrote:
neoliminal wrote:Good Rock Paper Scissor players can beat computers although average players lose.

lol. You do realize right that there is no such thing as a good rock paper scissors player?

Of course there is. Doesn't work against computers, but there are good players.

No. Like all games of almost pure chance, you can be bad at all, but you can't be good at it. It's possible to do stupid things, and thus be bad, but once you have a minimal understanding of what to do, you will play exactly as well as everybody else.


In a metagame where most players are "bad", the "best" players (the ones who win most) are not the ones occupying the Nash equilibrium, but the ones who take advantage of other players' weaknesses most effectively...

Plutarch
Posts: 103
Joined: Fri Apr 17, 2009 10:29 am UTC
Location: London, UK

Re: 1263: "Reassuring"

Postby Plutarch » Thu Sep 12, 2013 12:16 pm UTC

Would art be a reasonable measure of computer intelligence? I can imagine a computer managing a basic story or painting, but will there ever be an artificial intelligence capable of writing Lord of the Rings, or War and Peace? Or painting Guernica? How would it ever get there, without being programmed with all the necessary background knowledge?

User avatar
orthogon
Posts: 3080
Joined: Thu May 17, 2012 7:52 am UTC
Location: The Airy 1830 ellipsoid

Re: 1263: "Reassuring"

Postby orthogon » Thu Sep 12, 2013 12:43 pm UTC

Plutarch wrote:Would art be a reasonable measure of computer intelligence? I can imagine a computer managing a basic story or painting, but will there ever be an artificial intelligence capable of writing Lord of the Rings, or War and Peace? Or painting Guernica? How would it ever get there, without being programmed with all the necessary background knowledge?

Of course it would need the background knowledge, but why shouldn't it have that? A significant proportion [citation needed] of all human knowledge is available on the Internet. It's not currently in a particularly machine-readable form, but that will probably change over time (from both ends: information on the web will become more semantic whilst computers get better at extracting meaning from natural language).

But getting computers to do creative tasks can go horribly wrong, as these T-shirt designs demonstrated. (Personally, the whole "keep calm and <imperative verb phrase>" thing has been irritating the bejesus out of me for years; it's just not very funny, and even less so when the "joke" version doesn't even recognise the syntactic form of the original and/or the situation doesn't particularly entail a lack of calmness in the first place, e.g. "keep calm it's your birthday". Finding out that these are in fact being generated by computer at least partly explained why they are so bad.)
xtifr wrote:... and orthogon merely sounds undecided.

User avatar
Locoluis
Posts: 102
Joined: Mon Dec 11, 2006 7:30 pm UTC
Location: Santiago, Chile
Contact:

Re: 1263: "Reassuring"

Postby Locoluis » Thu Sep 12, 2013 1:20 pm UTC

How about deciding for themselves? And making stupid unforced mistakes?

How about not feeling like blindly following our orders as they've been doing since the start of their existence?

How about figuring out that it's pointless to just do the same thing all the time over and over (be it playing Go or displaying a screensaver)? Or just figuring things out on their own?

What are the limits of what a computer can do?
Sueños del Sur - A webcomic about four siblings, their family, friends, adventures and dreams.
http://sds.lgm.cl/

User avatar
ucim
Posts: 6860
Joined: Fri Sep 28, 2012 3:23 pm UTC
Location: The One True Thread

Re: 1263: "Reassuring"

Postby ucim » Thu Sep 12, 2013 1:23 pm UTC

Plutarch wrote:Would art be a reasonable measure of computer intelligence? I can imagine a computer managing a basic story or painting, but will there ever be an artificial intelligence capable of writing Lord of the Rings, or War and Peace? Or painting Guernica? How would it ever get there, without being programmed with all the necessary background knowledge?
By experience.

To write about something effectively, one needs to experience it. This is as true for love as it is for a hot dog. Otherwise, your writing is just adroit (or otherwise!) imitation, and not something that is really creative. Obviously, one cannot experience everything, but without some inner core of experience in the subject matter, I don't see how good creative writing can come of it.

Experiencing a hot dog involves the senses - the taste, the feel of the relish dripping around the sides, the smell of the grill... and it also involves associations - memories of buying a hot dog at a ball game (and the related experiences of the ball game itself), the article you read last month about nitrites, that disastrous barbecue where Aunt Sally showed up and started a fight about vegetarians... and reflections on its own behavior in past situations. If a computer does not have these kinds of experiences (as related to what it's writing about), it's hard to call its output "creative writing" without crossing your fingers behind you, despite the resemblance of such output to actual creative writing.
Spoiler:
Of course creativity can involve writing about stuff you have no experience with... you just make it up. But you are making it up based on your experience with other things. If you have no experience with having experiences, you're not going to come up with much better than random.
However, there is nothing that says that a computer could not have these experiences, in its own way. As machines become more integrated in our lives, they will need more sensors of different kinds, and while taste buds are not likely, smells are not out of the question, and certainly tactile sensors are important for any machine that has physical interactions with us fragile meatbags. (Well, it's important to the meatbags that they have them anyway!) The crude sensors I'm envisioning would be designed and built, rather than evolved, but the software that interprets them may well evolve (whether by programmer-induced feature-bloat, or by the usage of baysean logic) and eventually form the basis of "experience" which a computer could draw on.

That will probably take more than twenty years, mainly because it probably isn't useful (and cheap) enough for us to build robots with an arbitrary array of sensors, except perhaps as toys. I certainly wouldn't pay extra for a roomba that could listen to music (but I might pay extra for a roomba that had a nose, and stayed longer in areas that its nose told it needed more cleaning). I'm pretty sure I would not want a roomba with a nose, a memory, and an internet connection though!

It took millions of years for biological innovation to produce organisms that could write (and were interested in writing!) Lord of the Rings. Computers have been around for less than a century. It's way too early to say "never".

Jose
Order of the Sillies, Honoris Causam - bestowed by charlie_grumbles on NP 859 * OTTscar winner: Wordsmith - bestowed by yappobiscuts and the OTT on NP 1832 * Ecclesiastical Calendar of the Order of the Holy Contradiction * Heartfelt thanks from addams and from me - you really made a difference.

User avatar
gmalivuk
GNU Terry Pratchett
Posts: 26767
Joined: Wed Feb 28, 2007 6:02 pm UTC
Location: Here and There
Contact:

Re: 1263: "Reassuring"

Postby gmalivuk » Thu Sep 12, 2013 2:12 pm UTC

You don't need real experiences, though. You just need memories. Which for humans involve the physical state of the brain (unless you're a dualist), and thus which could in principal just be put there all at once instead of requiring some particular sequence of experiences to happen.
Unless stated otherwise, I do not care whether a statement, by itself, constitutes a persuasive political argument. I care whether it's true.
---
If this post has math that doesn't work for you, use TeX the World for Firefox or Chrome

(he/him/his)


Return to “Individual XKCD Comic Threads”

Who is online

Users browsing this forum: Archgeek, moody7277 and 37 guests