Bay 12 Games Forum
Finally... => General Discussion => Topic started by: Char13magne on May 04, 2010, 08:00:18 am
-
So, hi again guys!
My idea is thus;
humans have great calculating and computing abilities, and are even able to predict the outcomes of events in such narrowed down and focused things as, say, videogames. However, we lack both the memory and the observation skills to take note of and apply the effects of multiple variables from our surroundings into our mental computations, and this is why the ability to predict the outcomes of events in real life is such a rare and treasured trait. So, your thoughts gentlemen?
-
Wut?
Humans can predict video games, but can't predict the outcomes of real life? This is your idea? At the risk of being offensive... Wut?
Humans constantly predict the outcomes of events in real life. Its how we live our lives. Without the ability to predict the outcome of things, our lives would be incomprehensable strings of completely random events.
-
We predict things constantly. Thats exactly what we do. We put our food in cold places so it doesnt rot and hunt animals by predicting where they will run and setting up traps. Does a rabbit do that?
-
Many animals do that too. You dont require ellaborate abstract thoughts (which is what humans do) to predict stuff (even if it helps). For instance, predicting how a potential prey will bolt is essential to predators.
-
I'd be very scared if I couldn't predict things. Infact, mostly likely I'd be dead.
-
If I understand what the OP is trying to get at (and that's a big if), it's that we can predict narrowly modeled simulations or simple scenarios based on past experience. What we suck at is predicting complex and/or unprecedented situations. We can't predict "the future" (i.e. the sum total of billions of discrete events over a given time horizon).
To which, I'd have to say...duh? It's easy to predict something with two variables. As opposed to a complex geopolitical situation which literally has thousands of significant variables. And numerous potential outcomes.
-
I disagree completely. Making good predictions on narrow sets of data is exactly what computers are good at; humans are good at it too because it's not that hard. What living brains excel at is using past information and experiences, an incomplete and non-codified data set, and no strict instructions, and making all the predictions it wants about any complex subject. Of course accuracy falls with lack of information, and if you throw an oddball question at someone that they have no familiarity with, all they can say is they don't know. But predicting the product of a "calculation" without even knowing what the "calculation" is or needing to ascribe any hard rules or variables to it is something humans do all the time, about every subject. That's what intuition is.
Sure, our memory of past scenarios isn't always that great, but we are talking about a storage system based on active chemistry, if we even do know what memory is based on. We're still surprisingly good about remembering things, precisely because we don't have to enumerate everything in our entire experience to use it in a prediction. Present two people with a problem, whatever it is, and the one with more experience with similar problems will give a more accurate prediction, even if he can't precisely describe a single example he's basing that prediction on. We remember stuff deep in the intuitive brain a lot better than we're able to consciously express.
-
If anything people are too quick on the inductive draw, believing crazy things because of a few disparate data points.
My cows stopped giving milk because you're a witch, by the way.
-
I'd qualify that by saying that computers are far better at predicting things which can be mathematically modeled *and* which have accounted for most, if not all, of the significant variables. A computer will generally be a far more accurate and faster solution to plotting an artillery strike, for instance.
What we do better is "fuzzy" calculation, where we make a rough estimate to account for unknown variables based on past experience without necessarily being able to quantify or even explain why. More experience, better estimates.
In theory, a good enough AI could do the same thing, and faster. But modeling that heuristic loop that allows humans to do fuzzy learning is one of the core problems with AI design.
-
If I understand what the OP is trying to get at (and that's a big if), it's that we can predict narrowly modeled simulations or simple scenarios based on past experience. What we suck at is predicting complex and/or unprecedented situations. We can't predict "the future" (i.e. the sum total of billions of discrete events over a given time horizon).
This is more what I was getting at. Although I would like to add that in the short term, even our lack of some refinement in our more advanced organs that allow extreme perception can prevent us from predicting a proper outcome (an example that displays the opposite, Sherlock Holmes can predict future events because his senses are super-humanly acute, albeit he is fictional).
-
Humans can't predict the future, eh?
In other news, the sky is blue!
-
Humans can't predict the future, eh?
In other news, the sky is blue!
In this case, you've just missed the point, we can't predict complicated events well into the future.
-
Humans can't predict the future, eh?
In other news, the sky is blue!
In this case, you've just missed the point, we can't predict complicated events well into the future.
No, I get the point, it's just moot.
-
Humans can't predict the future, eh?
In other news, the sky is blue!
In this case, you've just missed the point, we can't predict complicated events well into the future.
No, I get the point, it's just moot.
My point, oh, alright, your opinion I guess, which I see is shared by the rest of the viewing community :-\. :-[
-
yes we can. Just check a weather forecast.
-
yes we can. Just check a weather forecast.
But are they always correct? I thought not.
-
This thread is on the thin, jagged edge of morphing into a global warming debate ...
... but you didn't hear that from me! ;)
On topic, now: the real question should be are we as good at modeling thought as we are at modeling ideas, and the answer for now is "no".
At least some kinds of complex prediction veer into np-hard territory, so what we're discussing is really an AI design problem.
I predict that methods for growing AI's will eventually get good enough we can grow an AI that will "think" better than a human, and I further speculate that at that point -- but not before -- at least some AI's will be able to consistently make np-hard predictions better than us. (The ability to copy those AI's makes it theoretically possible that almost all AI's could be better predictors.)
In the real world human-AI hybrid systems (tech-savvy managers with sophisticated data mining tools) are already outperforming either one separately.
Edit: and several studies have shown that even a monkey can out-perform a bad manager. :P
-
yes we can. Just check a weather forecast.
But are they always correct? I thought not.
Not always right /= Not right.
Just because we cannot calculate the future with 100% accuracy does not mean we cannot predict the future. It just means there is a chance we could be wrong, or the details are a little fuzzy. Considering that for a great many things we can predict the future in order to survive hostile environments to the degree that we have, I would say our ability to predict the future does have some amount of legitimacy.
-
This thread is on the thin, jagged edge of morphing into a global warming debate ...
... but you didn't hear that from me! ;)
On topic, now: the real question should be are we as good at modeling thought as we are at modeling ideas, and the answer for now is "no".
At least some kinds of complex prediction veer into np-hard territory, so what we're discussing is really an AI design problem.
I predict that methods for growing AI's will eventually get good enough we can grow an AI that will "think" better than a human, and I further speculate that at that point -- but not before -- at least some AI's will be able to consistently make np-hard predictions better than us. (The ability to copy those AI's makes it theoretically possible that almost all AI's could be better predictors.)
In the real world human-AI hybrid systems (tech-savvy managers with sophisticated data mining tools) are already outperforming either one separately.
Edit: and several studies have shown that even a monkey can out-perform a bad manager. :P
Why does everyone think now that just because I start a thread means I have to be trying to spark a debate? :P
-
yes we can. Just check a weather forecast.
But are they always correct? I thought not.
Not always right /= Not right.
Just because we cannot calculate the future with 100% accuracy does not mean we cannot predict the future. It just means there is a chance we could be wrong, or the details are a little fuzzy. Considering that for a great many things we can predict the future in order to survive hostile environments to the degree that we have, I would say our ability to predict the future does have some amount of legitimacy.
I never said humans were incapable of predicting into the future, just not good at it, oh, and weathermen use computers!!!
-
yes we can. Just check a weather forecast.
But are they always correct? I thought not.
Not always right /= Not right.
Just because we cannot calculate the future with 100% accuracy does not mean we cannot predict the future. It just means there is a chance we could be wrong, or the details are a little fuzzy. Considering that for a great many things we can predict the future in order to survive hostile environments to the degree that we have, I would say our ability to predict the future does have some amount of legitimacy.
I never said humans were incapable of predicting into the future, just not good at it, oh, and weathermen use computers!!!
You said it is a rare and treasured trait to be able to predict anything in our daily lives, implying that its something that is beyond the grasp of most mere mortals.
Why does everyone think now that just because I start a thread means I have to be trying to spark a debate?
You post threads every day that consist of "controversial topic, discuss!" This one only differs in that it's "I kinda thought of something, discuss!"
Also, double posting is bad form, you know this because people keep telling you.
-
yes we can. Just check a weather forecast.
But are they always correct? I thought not.
Being so literal is "such bad form" (to quote you a little bit, oh, and yes I'm being quite hypocritical right now with the literal part).
Not always right /= Not right.
Just because we cannot calculate the future with 100% accuracy does not mean we cannot predict the future. It just means there is a chance we could be wrong, or the details are a little fuzzy. Considering that for a great many things we can predict the future in order to survive hostile environments to the degree that we have, I would say our ability to predict the future does have some amount of legitimacy.
I never said humans were incapable of predicting into the future, just not good at it, oh, and weathermen use computers!!!
You said it is a rare and treasured trait to be able to predict anything in our daily lives, implying that its something that is beyond the grasp of most mere mortals.
Why does everyone think now that just because I start a thread means I have to be trying to spark a debate?
You post threads every day that consist of "controversial topic, discuss!" This one only differs in that it's "I kinda thought of something, discuss!"
Also, double posting is bad form, you know this because people keep telling you.
Being so literal on the forums is "such bad form" (to use your very words).
And I've only posted one, two topics at tops like this. Would you call "The Holy Roman Empire, neither Holy Nor Roman Nor An Empire," or "Wars Of The Roses, Yorkist Or Lancastrian" controversial?!?!?!?!?!?!?!?!?!? Also, my response to him in that case was more of a joke, but, hey, however you took it.
And regarding the second point you've made, I'm getting better aren't I? There aren't that many people who are perfect at this 100% of the time, that I can tell.
Oh, and don't tell me you don't see threads just like this all the time!
-
www.redorbit.com/news/science/126649/can_this_black_box_see_into_the_future/
Who needs to see into the future? I've got a random number generator for that.
-
Being so literal on the forums is "such bad form" (to use your very words).
And I've only posted one, two topics at tops like this. Would you call "The Holy Roman Empire, neither Holy Nor Roman Nor An Empire," or "Wars Of The Roses, Yorkist Or Lancastrian" controversial?!?!?!?!?!?!?!?!?!? Also, my response to him in that case was more of a joke, but, hey, however you took it.
And regarding the second point you've made, I'm getting better aren't I? There aren't that many people who are perfect at this 100% of the time, that I can tell.
Oh, and don't tell me you don't see threads just like this all the time!
It just seems to me like you've been on some sort of epic e-peen quest to get your post count as high as possible since you joined the forums. Sorry if im wrong, but it's kinda annoying.
Also: You're confusing being literal with quoting someone. Being literal is taking a statement that was meant to be figurative, like a metaphor, simile, or turn of phrase, and taking the words at face value, as opposed to their intended meaning. I just quoted you.
-
Oh, and don't tell me you don't see threads just like this all the time!
It just seems to me like you've been on some sort of epic e-peen quest to get your post count as high as possible since you joined the forums. Sorry if im wrong, but it's kinda annoying.
Sorry if I should, but I have no idea what that means. Anyone care to enlighten me?
-
Trying to gain popularity by making a lot of posts.
It doesn't work, by the way.
-
Trying to gain popularity by making a lot of posts.
It doesn't work, by the way.
Wouldn't that just get annoying? I see many flaws in the thought process behind that idea.
-
I ran some numbers in Visual Basic and it predicted that everyone will get way too tense, and someone will suggest that we just mellow out, and someone else will get angry about that.
I'm not sure.
-
I ran some numbers in Visual Basic and it predicted that everyone will get way too tense, and someone will suggest that we just mellow out, and someone else will get angry about that.
I'm not sure.
What's "Visual Basic"?
-
You guys need to just chill out.
-
Raaaaaaaaaaaeeeeeeggggg
-
Visual Basic (http://lmgtfy.com/?q=visual+basic+wikipedia) - predicting the future since it time-traveled back to 1883.
-
yes we can. Just check a weather forecast.
But are they always correct? I thought not.
Not always right /= Not right.
Just because we cannot calculate the future with 100% accuracy does not mean we cannot predict the future. It just means there is a chance we could be wrong, or the details are a little fuzzy. Considering that for a great many things we can predict the future in order to survive hostile environments to the degree that we have, I would say our ability to predict the future does have some amount of legitimacy.
I never said humans were incapable of predicting into the future, just not good at it, oh, and weathermen use computers!!!
The weather-computers give accurate forecasts out farther than people do.
However, at a certain point they are mere estimations, just usually better than what a person could give you.
I disagree completely. Making good predictions on narrow sets of data is exactly what computers are good at; humans are good at it too because it's not that hard. What living brains excel at is using past information and experiences, an incomplete and non-codified data set, and no strict instructions, and making all the predictions it wants about any complex subject. Of course accuracy falls with lack of information, and if you throw an oddball question at someone that they have no familiarity with, all they can say is they don't know. But predicting the product of a "calculation" without even knowing what the "calculation" is or needing to ascribe any hard rules or variables to it is something humans do all the time, about every subject. That's what intuition is.
Sure, our memory of past scenarios isn't always that great, but we are talking about a storage system based on active chemistry, if we even do know what memory is based on. We're still surprisingly good about remembering things, precisely because we don't have to enumerate everything in our entire experience to use it in a prediction. Present two people with a problem, whatever it is, and the one with more experience with similar problems will give a more accurate prediction, even if he can't precisely describe a single example he's basing that prediction on. We remember stuff deep in the intuitive brain a lot better than we're able to consciously express.
human brains are frequently better than a python interpretter!
speaking in non-exact terms of course.
...
-
You guys need to just chill out.
aaand....
Raaaaaaaaaaaeeeeeeggggg
bing.
-
You guys need to just chill out.
aaand....
Raaaaaaaaaaaeeeeeeggggg
bing.
I ... I mentioned this.
-
I wasn't really raeging.
-
I assumed.
-
It's difficult to truly raeg, you know.
-
yes we can. Just check a weather forecast.
But are they always correct? I thought not.
Not always right /= Not right.
Just because we cannot calculate the future with 100% accuracy does not mean we cannot predict the future. It just means there is a chance we could be wrong, or the details are a little fuzzy. Considering that for a great many things we can predict the future in order to survive hostile environments to the degree that we have, I would say our ability to predict the future does have some amount of legitimacy.
I never said humans were incapable of predicting into the future, just not good at it, oh, and weathermen use computers!!!
The weather-computers give accurate forecasts out farther than people do.
However, at a certain point they are mere estimations, just usually better than what a person could give you.
I disagree completely. Making good predictions on narrow sets of data is exactly what computers are good at; humans are good at it too because it's not that hard. What living brains excel at is using past information and experiences, an incomplete and non-codified data set, and no strict instructions, and making all the predictions it wants about any complex subject. Of course accuracy falls with lack of information, and if you throw an oddball question at someone that they have no familiarity with, all they can say is they don't know. But predicting the product of a "calculation" without even knowing what the "calculation" is or needing to ascribe any hard rules or variables to it is something humans do all the time, about every subject. That's what intuition is.
Sure, our memory of past scenarios isn't always that great, but we are talking about a storage system based on active chemistry, if we even do know what memory is based on. We're still surprisingly good about remembering things, precisely because we don't have to enumerate everything in our entire experience to use it in a prediction. Present two people with a problem, whatever it is, and the one with more experience with similar problems will give a more accurate prediction, even if he can't precisely describe a single example he's basing that prediction on. We remember stuff deep in the intuitive brain a lot better than we're able to consciously express.
human brains are frequently better than a python interpretter!
speaking in non-exact terms of course.
...
See Also: http://en.wikipedia.org/wiki/Artificial_neural_network
Our brains basicly just use a method of classification en-mass. Hence why it takes years of practice to make drawings which aren't just symbols representing that which we think we are drawing. Even then, you are focusing on the shapes, which themselves are symbols of what is seen.
-
It's difficult to truly raeg, you know.
Some would call it impossible.