Month: September 2022

EMERGING TECHNIQUES. "The robot observes its own gripper and corrects its motions in real time. It also exhibits interesting pre-grasp behaviors, like isolating a single object from a group. All of these behaviors emerged naturally from learning, rather than being programmed into the system." http://googleresearch.blogspot.com.es/2016/03/deep-learning-for-robots-learning-from.html View in LinkedIn

GENERAL LEARNING SKILLS for robots. Interesting video. https://lnkd.in/dpeU392 View in LinkedIn

Q: "How will AI reach a level, where computers do something that they were not programmed to do? Or asked differently, is the plot of I, Robot plausible according to computer science?" https://lnkd.in/dWhwFiV View in LinkedIn

A: "Yes. This is actually very plausible. People usually think of computer programs as something deterministic and predefined. This is true only for simple, sequential programs. But as computer science develops, our programs are becoming more and more unpredictable." https://lnkd.in/dWhwFiV View in LinkedIn

#1. UNPREDICTABILITY. "First reason for this is parallelization. Even simplest programs, when run on many processor cores, start to behave in non-deterministic way. The reason for this is that different cores compute with slightly different speeds and each time the sequence of computed tasks might differ." https://lnkd.in/dWhwFiV View in LinkedIn

#2. MYSTERY PACKAGES. "Second reason is machine learning. For many problems programmers just implement rough algorithm, and then drop large amount of "test cases", for computer to learn from. What is most important: programmers often do not know what are those cases, and there is often too much of them even to analyze by humans. As a result, no human knows exactly what computer learned." https://lnkd.in/dWhwFiV View in LinkedIn

#3. BRAIN-LIKE. "Third reason is emerging field of neuromorphic. Our computers are very inefficient in comparison to brains, partially because of their architecture. There are projects to modify computer architecture to be more like the brain, which result in further loss of control of what is going on inside. IBM Unveils a ‘Brain-Like’ Chip With 4,000 Processor Cores. All of this leads to computers that are no longer "programmed in some specific way". https://lnkd.in/dWhwFiV View in LinkedIn

FINAL OUTCOME. "All of this leads to computers that are no longer "programmed in some specific way". Instead, we are going towards computers that are taught to do something, but their behaviour is very complex and nobody really knows what are they doing." https://lnkd.in/dWhwFiV View in LinkedIn

FREEDOM. "A robot may be given a problem to which we don't necessarily pre-program a specific solution and it could come up with a solution. In that case it was programmed to make _A_ decision, just not specifically a pre-determined (pre-computed) decision." https://lnkd.in/dWhwFiV View in LinkedIn

SELF-AWARE BOT. "Scientists at Bielefeld University demonstrated the software they developed that gave a walking robot a simple form of consciousness (wherein it was able to overcome obstacles while traversing a course). Over the last few months, they’ve built upon this achievement, developing a software architecture based on artificial neural networks that will allow the bot to see itself as others see it; that is, programming that makes the robot self-aware." https://lnkd.in/dWhwFiV View in LinkedIn