IBM's Watson super computer

[DRSX]

So today I heard of this super computer called "Watson" I was reading a new article about having Watson run for president, this is a huge huge step for artificial intelligence and who knows maybe in 30-50 years all the politicians will be repalaced with super computers who can make better choices. It was already stated in 2014 that this machine can make better choices than a human based on the facts and the data it has access to.

"Watson received the first place prize of $1 million.[7]

Watson had access to 200 million pages of structured and unstructured content consuming four terabytes of disk storage[8] including the full text of Wikipedia,[9] but was not connected to the Internet during the game.[10][11] For each clue, Watson's three most probable responses were displayed on the television screen. Watson consistently outperformed its human opponents on the game's signaling device, but had trouble in a few categories, notably those having short clues containing only a few words."


What is really breathtaking is that they rely on Watson for cancer treatment options.

"In February 2013, IBM announced that Watson software system's first commercial application would be for utilization management decisions in lung cancer treatment at Memorial Sloan Kettering Cancer Center in conjunction with health insurance company WellPoint.[12] IBM Watson's former business chief Manoj Saxena says that 90% of nurses in the field who use Watson now follow its guidance.[13]"

The robot/super computer has also passed a self awareness test, it also has the capability of learning things.


https://www.youtube.com/watch?v=WFR3lOm_xhE

 

[nodnarbusn]

Fucccccccckkkkkkkkk THAT. Ill dip out before I have a robot overlord strait up. AI hasn't come very far as is, why the fuck would you immediately start working on a supreme overlord lol.

 

[DRSX]

Fucccccccckkkkkkkkk THAT. Ill dip out before I have a robot overlord strait up. AI hasn't come very far as is, why the fuck would you immediately start working on a supreme overlord lol.

Right not with the current politicians I'd support the robot no joke, and yeah AI hasn't come very far however it is progressing quite nicely and it is just a matter of time before AI takes over more things.

 

[RoundGift]

it's the end of humans !

 

[Xier0]

No surprise -- computers have dominated humans for awhile in areas such as chess. Watson is different. It just a program that finds out what the quiz question is intending to ask, then it basically does a google search of Wikipedia + many other sources (stored internally on a giant database). After it runs a few extra checks, it weighs the possible answers and will make an answer, or not. Basically, the struggle of the programmers is to make the computer smart enough to understand humans who have a dumb language.

Of course that's a really simplistic way to describe the program. If you want to listen to one of the designers explain it:

[YOUTUBE]3G2H3DZ8rNc[/YOUTUBE]

 

[MohtasaUnique]

Not to quote fiction as fact, but have you ever seen the movie iRobot? Do you remember the scene where the assistant bot chose to help Will Smith's character over the little girl in the other car because he had a slightly higher statistical chance of surviving, and a better impact on the community as a member of that society because of his status?

I believe putting AI in charge of a country would result in a lot of logical, yet unethical decisions much like this. I understand it's fiction, but that concept is a simple one; passing a bill solely based empirical data pulled from some statistics database is deeply a flawed and precarious method.

Additionally, the nature of politics isn't objective data, it's variance and diversity. If a room full of AI super computers were coming to a decision, it seems logical to assume all of the machines will reach the same conclusion as they would all essentially be vying for a statistical solution using the same source of information (internet, I'd assume), and there wouldn't be any point to debating or discussions. There wouldn't be politics. There would only be "problem solving", but the solutions may not be in everyone's best interest.


Having DeepBlue or Watson (or an equally powerful, but special-purposed machine tailored for political decision making) as an official consultant or reference in any debate may be beneficial however. But with our current technology, a government cannot be run effectively on AI. That's not to say AI won't eventually transcend the capability of human consciousness, but it's not anything we will be able to understand in the present day, and so there's no point to try and imagine that outcome.

 

[DRSX]

Not to quote fiction as fact, but have you ever seen the movie iRobot? Do you remember the scene where the assistant bot chose to help Will Smith's character over the little girl in the other car because he had a slightly higher statistical chance of surviving, and a better impact on the community as a member of that society because of his status?

I believe putting AI in charge of a country would result in a lot of logical, yet unethical decisions much like this. I understand it's fiction, but that concept is a simple one; passing a bill solely based empirical data pulled from some statistics database is deeply a flawed and precarious method.

Additionally, the nature of politics isn't objective data, it's variance and diversity. If a room full of AI super computers were coming to a decision, it seems logical to assume all of the machines will reach the same conclusion as they would all essentially be vying for a statistical solution using the same source of information (internet, I'd assume), and there wouldn't be any point to debating or discussions. There wouldn't be politics. There would only be "problem solving", but the solutions may not be in everyone's best interest.


Having DeepBlue or Watson (or an equally powerful, but special-purposed machine tailored for political decision making) as an official consultant or reference in any debate may be beneficial however. But with our current technology, a government cannot be run effectively on AI. That's not to say AI won't eventually transcend the capability of human consciousness, but it's not anything we will be able to understand in the present day, and so there's no point to try and imagine that outcome.

You're absolutely correct, however in 50 years I think robots/computers will take care of most jobs on this planet.

 

[kmjt]

Not to quote fiction as fact, but have you ever seen the movie iRobot? Do you remember the scene where the assistant bot chose to help Will Smith's character over the little girl in the other car because he had a slightly higher statistical chance of surviving, and a better impact on the community as a member of that society because of his status?

I believe putting AI in charge of a country would result in a lot of logical, yet unethical decisions much like this. I understand it's fiction, but that concept is a simple one; passing a bill solely based empirical data pulled from some statistics database is deeply a flawed and precarious method.

Additionally, the nature of politics isn't objective data, it's variance and diversity. If a room full of AI super computers were coming to a decision, it seems logical to assume all of the machines will reach the same conclusion as they would all essentially be vying for a statistical solution using the same source of information (internet, I'd assume), and there wouldn't be any point to debating or discussions. There wouldn't be politics. There would only be "problem solving", but the solutions may not be in everyone's best interest.

If the AI could be programmed to rescue someone based on their higher statistical chance of surviving/contribution to society, it could just as easily be programmed to detect age. So really it is just a matter of how the designer wants to morally program it. The problem would be that no one agrees on one set of morally correct decisions, as you mentioned.

In the future if any kind of significant AI is developed, I really hope it is an open source project and not just based on one person's view if what is right/wrong. Imagine someone like Hitler developing AI.

 

moved to tech

also: next step is to hack the president lmfao

 

[MohtasaUnique]

If the AI could be programmed to rescue someone based on their higher statistical chance of surviving/contribution to society, it could just as easily be programmed to detect age. So really it is just a matter of how the designer wants to morally program it. The problem would be that no one agrees on one set of morally correct decisions, as you mentioned.

In the future if any kind of significant AI is developed, I really hope it is an open source project and not just based on one person's view if what is right/wrong. Imagine someone like Hitler developing AI.

The point of autonomous AI isn't to program a robot to make a decision a certain way, but to give it the ability to evaluate the situation itself, and make a decision on criteria it sees as a best fit to the scenario. Telling it "rescues must be made based on age" isn't artificial intellegence, it's programmed logic

 

[kmjt]

The point of autonomous AI isn't to program a robot to make a decision a certain way, but to give it the ability to evaluate the situation itself, and make a decision on criteria it sees as a best fit to the scenario. Telling it "rescues must be made based on age" isn't artificial intellegence, it's programmed logic

The ability to evaluate the situation itself is programmed logic. Computers/robots are only as good as they are programmed to be. I don't see how a computer will ever be able to "evaluate" a situation in the sense that it is truly autonomous. How can a computer be autonomous when it struggles to generate true random numbers? Computers/robots can appear very autonomous but deep down there would always be programmed logic. Correct me if i'm wrong.

 

[MohtasaUnique]

The ability to evaluate the situation itself is programmed logic. Computers/robots are only as good as they are programmed to be. I don't see how a computer will ever be able to "evaluate" a situation in the sense that it is truly autonomous. How can a computer be autonomous when it struggles to generate true random numbers? Computers/robots can appear very autonomous but deep down there would always be programmed logic. Correct me if i'm wrong.

And that's why I'm saying, our current capability of AI is that it's only as good as the programmer made it. That model cannot possibly function in politics because it needs to be able to handle situations the programmer him/herself did not intend the computer to directly understand. A programmer cannot understand and program logic into all eventualities, and yet, how are humans able to adapt to new situations?

With that logic in mind, AI will have to be MORE autonomous than a human brain in order to function in politics. It will have to learn, grow, and reason, it will have to be able to make new memories, and make decisions based on that memory, it'll have to sympathize and empathize. Both of those concepts are currently understood to be human traits, and only JUST being understood to be prevalent in animals. They've not been programmable, and so robots cannot assume a role in which sympathy is a regular deference.

 

[Xier0]

With that logic in mind, AI will have to be MORE autonomous than a human brain in order to function in politics.

Politics has nothing to do with logic.

The actual realization of AI is basically being the gods who start the process of evolution in computers. Currently, things are programmed so that their intended purpose is to act predictably. As KMJT mentioned, trying to get a computer to give you a truly random number is even a challenge (which we faced when building gambling programs), so getting a computer to actually conceptualize new experiences and respond independently is quite a feat.

 

[DRSX]

Politics has nothing to do with logic.

The actual realization of AI is basically being the gods who start the process of evolution in computers. Currently, things are programmed so that their intended purpose is to act predictably. As KMJT mentioned, trying to get a computer to give you a truly random number is even a challenge (which we faced when building gambling programs), so getting a computer to actually conceptualize new experiences and respond independently is quite a feat.

The thing is IBM or Google who are developing AI right now have billions in their budget and they have very skilled engineers to figure shit out for them, nothing is impossible for these companies. I truly believe they will be able to achieve whatever they are trying to do. Will be interesting to see how it continues on developing tho.

 

[Xier0]

The thing is IBM or Google who are developing AI right now have billions in their budget and they have very skilled engineers to figure shit out for them, nothing is impossible for these companies. I truly believe they will be able to achieve whatever they are trying to do. Will be interesting to see how it continues on developing tho.

Sure, there are plenty of resources and a lot of demand for AI, but that isn't something that currently obstructs its development. True AI is essentially infusing life into a machine, and currently, we only know how to get machines to do what we tell them to. Computer language has no concepts, just commands.

 

[kmjt]

And that's why I'm saying, our current capability of AI is that it's only as good as the programmer made it. That model cannot possibly function in politics because it needs to be able to handle situations the programmer him/herself did not intend the computer to directly understand. A programmer cannot understand and program logic into all eventualities, and yet, how are humans able to adapt to new situations?

Agreed. No one person will know the best solutions to every problem out there. It would take a very large collaboration to create this decision set.

 

[MohtasaUnique]

Politics has nothing to do with logic.

The actual realization of AI is basically being the gods who start the process of evolution in computers. Currently, things are programmed so that their intended purpose is to act predictably. As KMJT mentioned, trying to get a computer to give you a truly random number is even a challenge (which we faced when building gambling programs), so getting a computer to actually conceptualize new experiences and respond independently is quite a feat.

I think you're misunderstanding me, logic in this case refers to how a computer arrives to its answer. Ands, ors, xors, if/then/else, etc.

And so yes. Politics cannot be decided with logic.............. that's my entire premise. That's why I believe computers cannot participate in politics, not with the current AI technology.

 

[kmjt]

I think you're misunderstanding me, logic in this case refers to how a computer arrives to its answer. Ands, ors, xors, if/then/else, etc.

Politics cannot be decided with logic..............

Highly improbable in any foreseeable future, definitely not impossible. Just because the discipline of political decisions and their future impact isn't modern, you never know for the future. Right now this AI would be impossible, but it definitely is not impossible in the future.

 

[Roulette]

FUCK holy shit. What have humans come to