3 ms·

One shouldn't use human concepts for "easy" vs "hard" when applied to AI capabilities. Machines aren't humans. Coal mining is hard for humans, but easy for a Ba

by est31 6y ago

One shouldn't use human concepts for "easy" vs "hard" when applied to AI capabilities. Machines aren't humans. Coal mining is hard for humans, but easy for a Bagger 288. Similarily, computers can do millions of floating point operations per second, where even a single operation would be very hard for humans, especially when larger numbers are involved. Doing the laundry, cleaning the toilet, etc. are all "simple" tasks but we don't have any humanoid robots yet that can do them. Also important to note: human capabilities are no natural constant. Once you reach human performance in some area, it's not guaranteed that you'll plateau, but instead likely that you'll outstrip human capabilities, especially in areas where there hasn't been billions of years of evolution led engineering like abstract thinking.

ChuckNorris89 6y ago
>Doing the laundry, cleaning the toilet, etc. are all "simple" tasks but we don't have any humanoid robots yet that can do them. That's because we have millions of unfortunate underpaid workers on the market with no better options that will do these chores cheaper than it costs to design, build and maintain a robot that can do such things. Hell, Atlas from Boston Dynamics can already do parkour so I'm sure they can program it to scrub toilets and do laundry but who would buy it for that when the minimum wage is what it is? The tech is nearly here, the business case isn't until the price comes orders of magnitude down.
- est31 6y ago
  This is what the 3rd place winner of the World Robot Summit 2018 looks like cleaning a toilet: https://www.youtube.com/watch?v=vx3QdaPD8bw https://www.youtube.com/watch?v=vx3QdaPD8bw Your point about the workers is true, but there are economies of scale here. I don't think it's going to be different in robots than in other areas. As for the human competition, we often had cases where technology has put humans out of their job. They provide a baseline price that you need to undercut. Today we might use robots to clean nuclear waste sites. Tomorrow we use mass produce them and put them into people's homes. As long as you don't make a larger mess at the toilet than Fukushima you should be fine! :)

by est31 6y ago

ChuckNorris89 6y ago

>Doing the laundry, cleaning the toilet, etc. are all "simple" tasks but we don't have any humanoid robots yet that can do them. That's because we have millions of unfortunate underpaid workers on the market with no better options that will do these chores cheaper than it costs to design, build and maintain a robot that can do such things. Hell, Atlas from Boston Dynamics can already do parkour so I'm sure they can program it to scrub toilets and do laundry but who would buy it for that when the minimum wage is what it is? The tech is nearly here, the business case isn't until the price comes orders of magnitude down.

est31 6y ago
This is what the 3rd place winner of the World Robot Summit 2018 looks like cleaning a toilet: https://www.youtube.com/watch?v=vx3QdaPD8bw https://www.youtube.com/watch?v=vx3QdaPD8bw Your point about the workers is true, but there are economies of scale here. I don't think it's going to be different in robots than in other areas. As for the human competition, we often had cases where technology has put humans out of their job. They provide a baseline price that you need to undercut. Today we might use robots to clean nuclear waste sites. Tomorrow we use mass produce them and put them into people's homes. As long as you don't make a larger mess at the toilet than Fukushima you should be fine! :)