Discussion in 'Intelligence & Machines' started by Hypercane, Jul 28, 2004.
I have the same thoughts. Please Register or Log in to view the hidden image!
Please Register or Log in to view the hidden image!
Log in or Sign up to hide all adverts.
Some of the human stuff is good to "copy", some not. We are not very good in many things. That's why we need things like AI.
Here is how I defined emotion for the purpose of my private AI development.
Self-awareness itself does not function as a goal generator.
No, they just need good algorithms for analyses and reasoning.
Asimov's 3 laws are just one mans opinion, my opinion is that you couldn't program an AI machine.
Yes, it is possible, but it is not guaranteed.
Here is a little story I wrote about a war between emotional, and emotionless, AI.
No humans were harmed in the pursuit of this conflict...
we all know what a problem data loss can be, what if the same happens to a robot, and he goes crazy?? What if the part which stores the rules, just melts?? In that case the rules would be useless.
Not unless the laws are carefully hardwired into their system.
U have all kinds of ppl, in this world, what if some wierdo, like me Please Register or Log in to view the hidden image! goes and just bangs on the part where the rules are hard coded, or breaks in (hacks) the part of the robot? Any E-lectronic system is vulnerable to bieng hacked today! (and tommorow)
Err.. Three laws?
How, exactly, can we have three universal laws and still let the pentagon develop those flesh ripping killer robots we all know they would love to have?
Will the next generation of Predator drones refuse any duties other than giving pleasure flights to orphans?
I don't see the implementation of the three laws anytime soon.
ill stay out of the discussion,
just going to say this, if we dont have those laws hardcoded, why should the robots?
Of course if a robot murders someone it should face punishment. (Or not)
Baal, If we are AI systems then we have hardcoded what our maker(s) wanted to hardcode. If we aren't then that's the answer. Punishment? Maybe you should start with a simple guide "How to punish BAD computers" and sell it with low-speed tickets to regular freaks and with a big hammer to advanced freaks.
It would suck pretty hard if our future was a 2 hour shoe commercial staring Will Smith.
Jiri, if you only knew how the human brain worked then you would understand that the only thing that could be hardcoded is for us to be the best.
btw, if you raise the my AI as you would raise your child then it would not be a criminal, it would respect human laws. I could tell you why, but i won't. I have gotten to a point now were it simply does not matter any longer. So, this is the last i will write.
Baal, the "the best" is meaningful only when associated with a well defined goal. Because of the variety of subjects and their preferences, there is no particular general meaning of the term. You have said in another thread that your AI can be coded on regular PC. Try to understand that systems which are just processing numbers cannot be called criminals and cannot be really punished (they just can be misused or broken). They don't like "10100101" more than "01101010". They simply do not care what you do to them. We are different because our pain/pleasure is more than a number. When writing software, we are trying to meaningfully associate computer's behavior with our real world and get some valuable results. It often works well because we understand some of the real world logic, but the machine is still just moving and modifying numbers based on certain rules. We make choices about which rules are used (which are hardcoded and which can be modified based on the processed data). When a computer virus knocks down some servers (or modifies/damages important data) and someone dies because certain data could not make it through then the author of the virus is responsible + possibly some other people who could prevent it if doing what they were supposed to do. But not the virus or the server itself. Note that there are many types of important medical procedures (including high-profile surgeries) and transactions being done which depend on functionality of various types of computer networks. Code can kill people but people make the choices (and mistakes). Mr. Rikard Svensson, I believe I know the real reason why you keep failing to explain various things here (on SF) and elsewhere. Please come back when you learn how to defend your thoughts, prove your claims and how to be honest. Juggling with big names, famous theories and big claims without reasonable explanations is the kind of stuff which used to impress my very first girlfriend ;-). We were 14 years old. Good luck on your way out of the pseudoscience. And take it easy, many scientists used to be in that "trap" at some point. And let me give you a little advice. If your primary goal is to develop AI then do not spend too much time trying to "correct" theories which are only indirectly related to your AI development. You could spend the rest of your life with that and never finish any decent AI project. Focus on developing a system which can evaluate, develop and correct many more theories than you could possibly focus on.
Is it those incorrect assumptions that make you the professional AI developer or is there some hidden reason?
Once again, you obviously believe that intelligence lies in hardware, do you also believe that it is your leg that is the intelligence behind your capability to walk?
The binary code is not intelligence, it however enables building a software interconnected with hardware that in the end can reproduce intelligence.
One day you will understand how incorrect you are. The hardware cannot be the criminal, just the software. Not the AI programming, but the information that it has gotten on its own.
That i seriously doubt since you clearly do not understand how the human brain functions (human or any other intelligent organic lifeform for that sake)
You are so keen on Artificial General Intelligence but you have no idea how the General Intelligence in this world works.
But i have already done as you wanted, i am not writing anything more here. Just defending myself, but i am not posting anything.
I only have a few seconds now so I'll just quickly address the first part of your response and I'll get to the rest at some point later.
No hidden reason. Unlike you, I can afford being honest and I'm trying to be very clear when I have time to do so. And, just to clarify, only a minor part of my current professional development is AI related. I currently do most of my AI stuff in my private R&D (which involves great freedom as well as significant limitations). Being a scientist is more about moving forward systematically than being exactly correct ATM. Scientific theories need to be useful, not necessarily 100% correct. I do not think I was wrong. For example, you are the best in being you, but what does it mean to me? Nothing.
huh?!, are you serious? cause then you are 0% correct.
You do not even understand what "the best" means then. You clearly do not understand even the most primitive functions of the human brain.
Who else can be Baal Zebul as well as you can be?
yes, of course i am the best at being myself, there is no competion.
the self does contain the parameters of matter but it is not alone a parameter of matter.
Best = excelling all others (in something). You have to specify that "something" to make such statement possibly meaningful to the others and even if you do so, it may not be truly meaningful to all the others because their reality may have practically nothing to do with that particular “something”. And of course no subject can be the best in everything because of the logical conflict. When pointing the stuff you consider incorrect, it would help if you can use the magic word “because” (or so).. Your note about the parameters is IMO irrelevant in this case.
Separate names with a comma.