Robot evolution

V1 & V2 were 70+ years ago, and more recently the modern style of drones have been used by armies for more than twenty years. But not yet by terrorists. It would’ve happened by now if it was ever going to.

Nah, no terrorist robots, it won’t happen.
Haven't there been terrorist drone attacks on an oil refinery in Saudia Arabia?
 
When he asked, "Can you put them there?"

Why didn't it say, "Yes".

or

"Yes, is that what you want me to do?"

Why did it proceed to do it?

If he said, "Can you jump out of the window?"
Would it immediately do that?"
 
When he asked, "Can you put them there?"

Why didn't it say, "Yes".

or

"Yes, is that what you want me to do?"

Why did it proceed to do it?

If he said, "Can you jump out of the window?"
Would it immediately do that?"
How many humans would say “yes” rather than just do it. And similarly, how many humans would jump out of the window?
Also, what window?
 
The real question is what happens when 10 people ask the robot to do something all at the same time? What do you see robot? So the robot describes the limited view of what the video is showing us. Problem is, the robot is facing an extremely large open area with plenty of stuff in its view, plus another robot standing in the background, and there are probably a lot of lights. It could have replied the way it did if it was asked to describe what was on the table or what was within 5 feet of it. To see how the robot really works, and not some staged show like a pop music video, show the robot a ceramic cup and a pitcher of water. Tell it to break the cup, and then pour water into the cup without spilling a drop. The show ends with the words that all these robots will need a digital platform to operate from. This robotic scheme will also require a lot of electrical power. When you're using your robot that you don't own, does it have a meter running like a taxi?
 
When he asked, "Can you put them there?"

Why didn't it say, "Yes".

or

"Yes, is that what you want me to do?"

Why did it proceed to do it?

If he said, "Can you jump out of the window?"
Would it immediately do that?"
As @Harpo said. The whole point is to get a robot to respond more like a human would. We are always making massively ambiguous requests of each other and we interpret those requests based on context which is exactly what this robot did. "Can you put them there" is exactly what many people would ask instead of saying 'put them there'. The robot interpreted it to be that, which in this context was correct, and that in itself is remarkable. Had they asked 'can you jump out of the window' I expect it would have answered differently. Actually it probably would have said it didn't understand as I doubt it has been trained to respond to such a question.

The real question is what happens when 10 people ask the robot to do something all at the same time?
I don't think that's the real question at all. What would a human do in those circumstances? Probably answer just one and I expect this robot would do the same.

show the robot a ceramic cup and a pitcher of water. Tell it to break the cup, and then pour water into the cup without spilling a drop.
Why would you train a robot to respond to a meaningless question. Would you train a waiter to respond to such a question? It's useless. They are trying to get robots to do useful stuff (and, I'd say, succeeding). And of course it's going to consume lots of power. How much power did the early computers consume that were less powerful than a modern pocket calculator today? This is still very early days efficiency comes later.


I wonder sometimes what people expect. Do you expect the early attempts at robotics to birth R. Daneel Olivaw in their first iterations? Give them a break; what they are achieving now is remarkable and in ten years time it will be even more so.
 
They are demonstrating the software for investors to invest in, that's the AI part. I expect it to be a dog and pony show completely scripted to make sure the investors will want to invest in it today. How did the robot know to only say that it only saw the items on the table? There are clearly many items to be seen in the direction the root is facing. Why did the video show what the robot could see? Was that a mistake? Bad editing? The robot couldn't possibly know the guy was only talking about the items on the table unless it was programmed to do so. What it did I bet there are plenty of high school kids in robotic courses able to do the same things.
 
They are demonstrating the software for investors to invest in, that's the AI part. I expect it to be a dog and pony show completely scripted to make sure the investors will want to invest in it today. How did the robot know to only say that it only saw the items on the table? There are clearly many items to be seen in the direction the root is facing. Why did the video show what the robot could see? Was that a mistake? Bad editing? The robot couldn't possibly know the guy was only talking about the items on the table unless it was programmed to do so. What it did I bet there are plenty of high school kids in robotic courses able to do the same things.
I would have to disagree, it probably reasoned that the area of interest was the area that it and the human were both looking at. IE the area between them. Also the video was not a promotional video. Unless I'm completely mistaken this video is produced, along with a number of others about other AIs, by an independent v logger.

Having said that it's also perfectly possible that it was sensibly programmed to only pay attention to objects within it's reach. And I'd also say that if I asked a human what they can see in front of them, I'd probably not expect them to answer that they can see a mountain on the horizon!
 
Last edited:
564A35DD-30DC-4924-BF4E-560FADDAFB83.jpeg


 

Similar threads


Back
Top