Feedback learning and generalization in robotics takes a step forward.

There is an interesting robot created by Cornell that functions based on feedback from humans and then adds generalization on its own documented here. This is an important step as we continue to move robots into the service sector and replace humans doing the jobs.So what should we do with all the checkout folk, waiters, waitresses, counter clerks and other folk whose function gets surplussed?

But many displaced workers did not make the transition. The development of trainable robots that can learn and programmable robots that simply repeat the same task is significant. The generalized learning robot and also move on to the next soon to be defunct position as well.Edited by jar, : appalin spallin

Why would you hear of mass deaths? People die constantly. People also lost homes, families were split apart, pensions disappeared, health care was lost and so sickness became the norm.The advantage humans have had is that we were the generalists, we had the advantage of being generalists as opposed to specialists and so learned new tasks. The folk working on the line became the folk stocking shelves.How far can that go on though? When the machines are the generalists but lower cost generalists that do not require fringe benefits what is the fallback position?Gross National Product may increase while per household income continues to decrease and the gap between rich and poor continues to grow. How far can that go without serious consequences?

Did you read the article?The point is that the Cornell robot is NOT simply programmed but rather taught to learn, to take correction from the people in real time and to then be able to apply what is learned to apply to new situations in a generalized fashion.I also spent some time in the food service industry and maybe we can swap tales sometime, but start here. Think back to how you learned your jobs, then compare that to what is in the article.

So how can we avoid or minimize hunger, strife, riots, class warfare, revolution?Will the pain of hunger, strife, riots, class warfare, revolution be necessary to get the US to adapt to the change?

The point of the article is that the robot does not operate only from instructions.

The robot learns to respond the same way you learned.

The important difference in this case IMHO is that the Cornell (and several other current projects) is a change from the pre-programming scenario to learning by doing.One example introduced was the tasks of a waiter or waitress in the food service industry. Well, been there and seen it done and guess what; waiters and waitresses learn by doing. They make mistakes and get corrected. That is exactly how the Cornell robot in the OP learns. The difference is that the robot learns almost immediately and can then generalize from that learning experience AND it don't forget.

Known fact.The robots all look prettier at closing time.

Once robots are self-learning, what programming would be needed?