I don't think it is lying. It's just not reporting food.
Lying would depend on the understanding that it would always report any and all food it finds. Given the same scenario, if it's 'understood' that the robot would only report if it finds a certain amount, or that it just doesn't report at all, it's not lying. They don't have agreements with other robots, they're only follow their programming. They're not lying.
And besides, this is no more lying than a virtual cd-rom drive is lying to an operating system, or a modchip is lying to a console.
It actually is lying. If you read the report, the robots produce light randomly, as more and more find a food source the amount of light grows in in that area. The robots learned to stop producing light when they found food.
The point is that it's run by a neural net. The researchers didn't just tell it "stop flashing the light". They gave it a goal to accomplish, and let the neural net sort out how to accomplish it. As it happens the robots figured out that not reporting the find was the best way to do it, so it modified it's existing programming appropriately. It's true that this isn't quite "lying" as we know it, but it is a bit more sophisticated than a modchip or a virtual drive. It's a small step, but the point is it's one we can build on as we continue to design more complex AIs.
The Rip is the latest addition to the Boogie eWriter line, devices that let you scribble notes and drawings and can be wiped away with the press of a button.
The most commented posts on Engadget over the past 24 hours.
Now that we've thrown 'em off the trail, use the form below to get in touch with the people at Engadget. Please fill in all of the required fields because they're required.
I don't think it is lying. It's just not reporting food.
Lying would depend on the understanding that it would always report any and all food it finds. Given the same scenario, if it's 'understood' that the robot would only report if it finds a certain amount, or that it just doesn't report at all, it's not lying. They don't have agreements with other robots, they're only follow their programming. They're not lying.
And besides, this is no more lying than a virtual cd-rom drive is lying to an operating system, or a modchip is lying to a console.
It actually is lying. If you read the report, the robots produce light randomly, as more and more find a food source the amount of light grows in in that area. The robots learned to stop producing light when they found food.
But it says in the article that the robots had signed contracts with the other robots. Probably forged then.
The point is that it's run by a neural net. The researchers didn't just tell it "stop flashing the light". They gave it a goal to accomplish, and let the neural net sort out how to accomplish it. As it happens the robots figured out that not reporting the find was the best way to do it, so it modified it's existing programming appropriately. It's true that this isn't quite "lying" as we know it, but it is a bit more sophisticated than a modchip or a virtual drive. It's a small step, but the point is it's one we can build on as we continue to design more complex AIs.