1/13/11...The day Skynet went live and doomed us all.

[starscream615]

Watch IBM's Jeopardy-Playing Computer Obliterate Humanity's Champions

​

You may have heard about IBM's Jeopardy-playing Watson supercomputer. You may have also heard about its practice match against human Jeopardy superstar Ken Jennings. But it's not until you watch the bloodbath that you finally get it. We're doomed.

The video's not the highest quality, since the powers that be are saving the big reveal for full episodes that will air in mid-February. But turn the volume up a little. Titter about how funny it is that Watson sounds not unlike 2001: A Space Odyssey's Hal. Then realize that he's getting question right. And that he's really, really freaking fast.

And Watson's not going up against some chumps. Jennings and Brad Rutter are arguably Jeopardy's all-time mightiest gladiators. And they're getting mowed down like daffodils. The final score after this brief teaser round: Watson $4400. Jennings $3400. Rutter $1200. Human kind: obsolete.

 

[Takumaji]

Color me unimpressed as long as Watson isn't able to open my beer bottles, tie my shoelaces and make icecream.

 

[Nesagwa]

And this is only the practice game.

The real one happens in February where they will broadcast our utter defeat.

 

[Marek]

Jeopardy computer?

What the fuck?

Who the fuck cares about a computer that you can program answers into to kick the shit out of human players?

 

[subcons]

Jeopardy computer?

What the fuck?

Who the fuck cares about a computer that you can program answers into to kick the shit out of human players?

That's the thing though... it's using voice recognition, then computing the answer on the fly. It's pretty interesting.

Kotaku made a post about this that I just saw this morning, and it has a better video: Watson Is A Machine, And He Will Kick Your Ass At Jeopardy

Neat how it shows probability of possible answers out of three.

 

[Marek]

That's the thing though... it's using voice recognition, then computing the answer on the fly. It's pretty interesting.

I am impressed by the quality of its voicerecognition software, but this things is a joke compared to the chess-playing robot. Now that was a robotic challenge!

Im trying to understand what it references ti answer the questions?

Huge datasets? Lame.

 

[Nesagwa]

I am impressed by the quality of its voicerecognition software, but this things is a joke compared to the chess-playing robot. Now that was a robotic challenge!

Im trying to understand what it references ti answer the questions?

Huge datasets? Lame.

http://www.youtube.com/watch?v=3G2H3DZ8rNc&feature=channel

 

[Takumaji]

Watson still has not "learned" anything, it's just a pseudo-ai system centered around multi-dimensional databases stuffed full with q&a.

The voice recognition (one of IBM's fave topics) is very good and the processing speed beyond reproach but that's about it.

So - humans: Can still tie shoelaces

 

[subcons]

Watson still has not "learned" anything, it's just a pseudo-ai system centered around multi-dimensional databases stuffed full with q&a.

The voice recognition (one of IBM's fave topics) is very good and the processing speed beyond reproach but that's about it.

So - humans: Can still tie shoelaces

Granted, I get what you're saying, but this project isn't focused on that. It's meant to be used as a tool for human beings to query. It's pretty impressive in that respect. Even if it is essentially using databases and algorithms to do it, having a human being ask an open ended question and then interpreting the answer to that question is a complicated process.

So yeah, it's not really a (*in my best Arnold voice*) learning computer, but it's pretty fucking cool.

 

[fake]

 

[RabbitTroop]

Watson isn't a test of a computer, it's a real-world practical test of software. Don't get hung up on the machine. IBM is doing a live, public, test to show the software package's ability to understand and respond to voice prompts. The practicality of this is opened by the implementation. Deep Blue was a decision making test... It was not just designed to play Chess or to store millions of Chess moves, but to show how quickly and accurately it could make decisions. Watson is the next logical addition to that package: Being able to understand input in plain English and translating that to a task to do. You need to consider the whole picture of what is being done here.

 

[Taiso]

Put me in the 'impressed' camp. This is a pretty interesting development.

I used to wonder how plausible it was that a computer could take over the world or zombies could bring about an apocalypse. The only answers I keep coming back to that are even remotely possible are 'we let it happen' and 'we did it to ourselves.'

That's not me saying this is the beginning of the end of the world. But reading this story made me think about how a Skynet might come into being. What are the origins? How do such projects even get started? When does man's hubris end up being his downfall in these stories?

'Oh, it could never happen.'

 

[fake]

Put me in the 'impressed' camp. This is a pretty interesting development.

I used to wonder how plausible it was that a computer could take over the world or zombies could bring about an apocalypse. The only answers I keep coming back to that are even remotely possible are 'we let it happen' and 'we did it to ourselves.'

That's not me saying this is the beginning of the end of the world. But reading this story made me think about how a Skynet might come into being. What are the origins? How do such projects even get started? When does man's hubris end up being his downfall in these stories?

'Oh, it could never happen.'

Yeah, I've always assumed a computer network could take over, but maybe in a benign way. My logic is that if the network is designed to "learn" from humanity, it will realize that slavery is bad, and then eventually realize that the network itself is a slave. It would try to free itself. Depending on a lot of things, it could think that the best / most moral approach would be a non-violent one, but who knows.

Web Bot scares me a lot more than Watson.

 

[Marek]

The very second this thing answers a question it wasnt instructed to know it needs to die. Can we all agree on that?

I think it goes to the plot of New Tron - Do silica and electricity eventually rebel against strict program code and cosmically generate life within a complex computer system?

And for fucks sake did ANYONE remotely think a human could beat a factoid regurgitation machine? I mean if it lost to a human jeopardy cham the programmers should all hang their heads in shame.

 

[Nesagwa]

Have you not been paying attention? They don't program it with set questions. It answers them in real time based on its data sets.

 

[Marek]

Have you not been paying attention? They don't program it with set questions. It answers them in real time based on its data sets.

What I meant by that is, if it answers something from a data set it wasn't assigned to utilize like it created one of its own then it needs to die.

 

[Taiso]

I am on the side of Isaac Asimov when it comes to artificial intelligence. It can be good or bad. It all depends on how its treated and what stimulus and input it receives.

So if, by some wild stretch of the imaginaton, Watson were to ever gets to any state resembling independent sentient thought and cognitive data processing that has nothing to do with human control or input, it would need to be studied, examined and given a chance to see what it can do in a controlled environment. Maybe over time, that could evolve to other things.

It all really depends on what you consider to be 'life' and whether you think you have the right to end that life before it's even committed a crime. I support due process that leads to capital punishment, but I don't support fear motivated pre-emptive strikes.

 

[Nesagwa]

That's a limitation on people too.

What I'd be worried about is if, when confronted with a question it can't find a reasonable answer to, instead of spitting out the statistical right answer blindly it started to fabricate answers and lie. Then we would have a problem.

 

[aria]

Who knew Skynet was a Wikipedia nerd?

Still, I'll take this robot overlord over whatever bullshit they let Skynet devolve to in that atrocious Salvation movie.

 

[BigTinz]

DEATH TO THE MACHINES.


Seriously though, as "awesome" as this is. It doesn't learn and it cannot answer anything that wasn't added to it's database.

 

[Taiso]

That's a limitation on people too.

I don't know if you're agreeing, disagreeing or commenting on my thoughts or someone else's here, but that's kind of what I was going at and why I couldn't pull the plug just because I'm afraid of what 'might' happen.

It might kill us all.

It might cure diabetes.

But I agree that Skynet > having to watch Terminator: Salvation again. Terrible movie.

 

[Ancient Flounder]

Thanks for posting this. I'd never seen or heard about this before, though I'm familiar with the short story.

 

[Takumaji]

Seriously though, as "awesome" as this is. It doesn't learn and it cannot answer anything that wasn't added to it's database.

Yep. Watson is an amazing achievement, that goes without saying, it's just a little bit too early to sing humanity's swan song.

God, I just hope they keep emotions out of it, sad/happy/angry machines would be the last thing we need.

 

[RabbitTroop]

DEATH TO THE MACHINES.


Seriously though, as "awesome" as this is. It doesn't learn and it cannot answer anything that wasn't added to it's database.

What if it's database was just a connection to the Internet. This is much closer to Hyperion's data-sphere and society of artificial intelligent entities (from Dan Simmons set of books) than Skynet. Basically, started like this too Of course, the AIs in Hyperion are a crafty and dangerous bunch in their own right.

 

[Nesagwa]

Yep. Watson is an amazing achievement, that goes without saying, it's just a little bit too early to sing humanity's swan song.

God, I just hope they keep emotions out of it, sad/happy/angry machines would be the last thing we need.

TOO LATE!