Artificial Intelligence: Philosophy Forums

Philosophy Forums » Philosophy of Science » Artificial Intelligence » Artificial Intelligence

Artificial Intelligence
Commonly thought of as emotionless...I beg to differ

Page: 1 2

Artificial Intelligence

J.D.
Student

Usergroup: Members
Joined: Apr 13, 2008
Total Topics: 5
Total Posts: 73

Posted 09/08/08 - 12:08 PM: Subject: Artificial Intelligence	#1
So I'm no scientist, I'll say that up front. I'm just another guy that's fascinated with the most intriguing portions of life. I was just thinking the other day about how artificial intelligence might work. The fundamental issue is understanding. If a computer cannot understand, it cannot comprehend and as a result will fail to make relevant conversation or thought process. I could be wrong, but as a child, I learned and understood things by positive and negative feelings toward them. Among the first things we learn, is what tastes bad and what tastes good. We feel negatively about the bad, and therefore learn to avoid it. It is all about how an experience would make you feel, (regardless of its moral labels.) So, is not the fundamental issue of understanding, emotions? At the very least, positive and negative, if nothing else. Edited by J.D. on 10/17/08 - 09:31 PM. Reason: irregardless isn't a word >.<
"I am, I exist, is necessarily true whenever it is put forward by me or conceived in my mind." -René Descartes, Cogito ergo sum

SparrowShadow
Aspirant

Usergroup: Members
Joined: Jan 26, 2008
Total Topics: 1
Total Posts: 20

Posted 09/08/08 - 12:31 PM:	#2
I have worked with Daisy by Greg Leedberg, it may even still be out there (Ver 1.1). Anyhow my physics instructor attempted a turing test on it after i'd talked to it about three months, interestingly he balked when it finally concluded, and repeated to him over and over "You depend on others to exist."

swstephe
Tenured Poster

Usergroup: Moderators
Joined: Apr 20, 2006
Location: borneo island
Total Topics: 22
Total Posts: 2150

Posted 09/08/08 - 05:06 PM:	#3
I agree, having dabbled in AI myself. There seems to be an implicit assumption by the general public that AI would have to be without emotions, and the secondary assumption that if an AI had emotions, they would have to function in exactly the same way as human emotions. One way of looking at emotions is to consider them a weighing procedure against potential courses of actions toward a solution. The purpose of emotions is to continuously guide our actions and provide us with motivation. So far, I think AI has only scratched the surface of something that is definitely capable of being modeled. I've been playing with "constraint logic programming", where, instead of programming something, you give it a set of constraints, a goal and a set of rules to satisfy the goal. Sometimes the software will appear to exploit flaws in the rules, (we consider it "cheating", but it is simply that the rules were ambiguous). From the other perspective, I think that any time a program starts to demonstrate motives, (toward reaching a goal), we attribute "emotions" to the program. The missing piece is for a program to demonstrate motivation in terms that social animals understand.
Ethics is the measuring of morality. Morality is the measuring of good. Good is the measuring of benefit. Benefit is the measure of values.

DJPavel
Assistant Professor

Usergroup: Members
Joined: Oct 02, 2007
Total Topics: 8
Total Posts: 394

Posted 09/08/08 - 06:57 PM:	#4
I think it depends on what you mean by "Artificial Intelligence". If you take it to be something like Cartesian reason, i.e. some absolute rules of thought "out there" independent of the substrate in which the "brain" is implemented, then yes, we can certainly achieve AI without emotions. If, however, AI is defined in terms of "human cognition", then I completely agree - you can't disregard the emotional component. The brain damage patients (in the limbic system) and autistic children clearly demonstrate that. DJP

Budapest
Aspirant

Usergroup: Members
Joined: Jul 14, 2008
Location: France
Total Topics: 2
Total Posts: 21

Posted 09/10/08 - 02:10 PM:	#5
Hello, JD. There are some fundamental reasons why it appears to be impossible to imitate thought/consciousness on a digital computer (known as strong AI). The most obvious reason is that at our present level of understanding the thing we call thought/consciousness cannot be explained mathematically (and thus cannot be expressed algorithmically) and does not appear to work in a determistic manner; although you might have a hard job making your mind up about this. It's wondefully ironic that there are thousands of people all around the world trying to create strong AI, yet not a single one of them could define what 'intelligence' or 'consciousness' is (they're trying to create something they can't even define). With regard to emotion - which of course is a core component of the thing we call 'thought' - programmers hate (ha!) the emotion thang because it takes a whole lot of programming and one day the universe will end. Apart from the lack of mathematical and intellectual definition of 'thought', there are some other very strong reasons why it doesn't look like it can be done on a digital computer. But I think that's enough for now!
http://www.spiderbomb.com/burg42

swstephe
Tenured Poster

Usergroup: Moderators
Joined: Apr 20, 2006
Location: borneo island
Total Topics: 22
Total Posts: 2150

Posted 09/10/08 - 08:05 PM:	#6
Budapest wrote: The most obvious reason is that at our present level of understanding the thing we call thought/consciousness cannot be explained mathematically (and thus cannot be expressed algorithmically) and does not appear to work in a determistic manner; although you might have a hard job making your mind up about this. Just read "Consciousness Explained" by Daniel Dennet. The reason consciousness can't be explained mathematically is because we really don't know what consciousness is, other than some emotionally invested subjective experience. If you can define any model in terms of prediction and consequences, then you can create a model which can be programmed, (and it is either deterministic or probabilistic by definition). Dennet describes some sample prototypes, (based on descriptions by Popper and Skinner), which can definitely be modeled. But isn't it a common trap -- to expect AI to be "human consciousness" in every detail, rather than just something that performs the same function, or to interact with humans in a way that is indistinguishable from a human agent, (the Turing test)? Its like the old analogy: expecting an AI to be actually conscious is like expecting shoes to be feet. Budapest wrote: It's wondefully ironic that there are thousands of people all around the world trying to create strong AI, yet not a single one of them could define what 'intelligence' or 'consciousness' is (they're trying to create something they can't even define). That's not "ironic", that's science and/or technology. I don't know of any researchers trying to create strong AI. Most AI researchers focus on a certain aspect or function with the expectation that by trying to model that function, they will learn about how that function really operates. It should be wonderfully hopeful that we can perform so many tasks far better than any human brain despite not having a complete image of the human brain. Budapest wrote: With regard to emotion - which of course is a core component of the thing we call 'thought' - programmers hate (ha!) the emotion thang because it takes a whole lot of programming and one day the universe will end. Is that an "appeal to emotion" fallacy? Emotions are actually quite well understood, compared to consciousness. There are a finite set of chemicals that the brain releases and a fairly small region of the brain that gets activated which serves as a valuation guide. I'm a programmer, and I would love to model some emotions some day. There are already many models of emotions out there and have been since the late 60's. Just because you haven't heard of it, (or can't find it on Google), doesn't mean it doesn't exist. Budapest wrote: Apart from the lack of mathematical and intellectual definition of 'thought', there are some other very strong reasons why it doesn't look like it can be done on a digital computer. But I think that's enough for now! Be careful of predicting something based on lack of evidence in your current experience. A few centuries ago, the things we do today may have seemed impossible. A lot of people are on record as saying that some things were impossible -- that we would never travel faster than 60 mph without being crushed by the acceleration, that we would never fly, that we would never place a man on the moon, etc. How can you know that unless you understand not only the entire subject of AI, (which is very rare), as well as all developments that are going to happen in a rapidly developing arena with thousands of researchers. The AI field also has its share of surprises. One group has recently been able to create a system which can accurately translate languages without any semantic knowledge of the languages involved. It threw out 50 years of research in favor of something that should have been obvious -- but it was possible.
Ethics is the measuring of morality. Morality is the measuring of good. Good is the measuring of benefit. Benefit is the measure of values.

wuliheron
Tenured Poster

Usergroup: Members
Joined: Jun 02, 2003
Location: Newport News, Va
Total Topics: 10
Total Posts: 3157

Posted 09/11/08 - 01:18 AM:	#7
AI is not synonymous with consciousness. There are perfectly useful programs that have passed the Turing test and I suppose at some point in the future there will be others that will be able to fool your average user. Think of a search engine that asks you clarifying questions for example. That said, I do agree that emotions are integral to consciousness. You might find the work of Antonio Demasio interesting.

Budapest
Aspirant

Usergroup: Members
Joined: Jul 14, 2008
Location: France
Total Topics: 2
Total Posts: 21

Posted 09/11/08 - 12:32 PM:	#8
swstephe you said: "Emotions are actually quite well understood, compared to consciousness..." Well, perhaps you'd like to properly explain them then, as opposed to your previous extremely vague definition? (and please don't direct me to books written by quacks) swstephe you said: "But isn't it a common trap -- to expect AI to be "human consciousness" in every detail, rather than just something that performs the same function". The only example, the only working model we have of sentient beings is ourselves: biological consciousness. If programmers want to create 'some other kind of consciousness' that's fine. Methinks you'll find it harder than working out the square root of two, but if 'other consciousness' can be done it would be fascinating to see what kind of entity it is, and whether we could communicate with it in any way, and whether it could do anything useful. With regard to "human consciousness in every detail", that opens up a whole can of worms! Our thought processes are so imtimately connected to our physical body, and our senses/the world around us, and (another tough one to define) the state of 'being' (which includes things like sleep, reproduction and nourishment - not necessarily in that order) that it's not difficult to wonder if you can separate them and still have 'thought'? I'm quite happy for AI folks to clutch tightly to their stone axes (aka digital computers) because I'm fairly sure they'll never be able to imitate 'thought' (in the homo-sapien/biological sense) - because if they did manage to create true machine intelligence it would almost certainly be used for ever more efficient methods of slaughter. And we don't want that now, do we. wuliheron, if you read my previous post carefully you'll see that I'm only talking about strong AI (ie, a machine with a near-human level of consciousness), not AI in a more general sense, which is already here and doing useful things, particularly for the military.
http://www.spiderbomb.com/burg42

Pongobongo
Initiate

Usergroup: Members
Joined: Sep 17, 2008
Total Topics: 0
Total Posts: 2

Posted 09/17/08 - 02:55 PM:	#9
Would it be appropriate to create a new term then for re-creating so-called "human consciousness"? I personally much prefer "synthetic consciousness". This I believe to be what most people consider AI to be. Strong AI seems pointless. Why would we want to create beings with "near-human" level consciousness? Current "weak" AI is just fine for that. Otherwise nothing less than human-level is acceptable. Is a self-aware, self-controlling, self-motivating machine anything less than fully-intelligent? I think not. Let us not confuse the terms intelligent and smart. Everyone wants "human-like" robots, when what they're really asking for are "human adult-level" robots. Why not create a sort of robot "baby" and teach it from there? Don't try to force our human anatomy and way of thinking on it, let it figure itself out the way we do. Start with basic programing, teach it, and let it re-program itself. There is only one kind of "intelligence", real intelligence. Anything less is fake intelligence (no matter how human-like it may seem, it's still "pre-programmed").

Budapest
Aspirant

Usergroup: Members
Joined: Jul 14, 2008
Location: France
Total Topics: 2
Total Posts: 21

Posted 09/17/08 - 03:15 PM:	#10
Pongobongo. I prefer the term 'non-biological intelligence' (NBI) because AI is too closely associated with the digital computer. Almost the entire point of 'strong AI' is about ever more efficient methods of slaughter. Certain countries have poured literally trillions of dollars into AI research over the last four decades. Just about all AI research is funded by the military. There's no such thing as 'artificial intelligence' because if anyone is capable of creating an intelligence/consciousness it will be anything other than 'artificial'.
http://www.spiderbomb.com/burg42

Page: 1 2

You don't have permission to post.

Please login or register.