Focusing on the theme of field based technologies; I took opportunity to attend a panel discussion on Robotics by the Royal Society at the Science Museum in London, hosted by science broadcaster and author Dr Adam Rutherford .The panel specifically explored the future possibilities for intelligent robots, what they will do, how we will use them, and how they might use us.
The inclusion of philosophy, ethics, moral responsibility into any type of computer code was discussed, as perhaps impractical, simply because a computer is not morally culpable, making machine to machine interactions biased based of the type of A.I used. So the idea that one system can have all the answers without human involvement makes A.I artificial. The need to clean biased human involvement within data sets was also mentioned. The polemic cases for A.I discussed, as was the singularity theory, where understanding the moral and ethical obligation by the system, manufacturer and user can help decide where actual responsible accountability might exist e.g. if an exoskeleton was used instead of a wheelchair, and the user experienced malfunction – who would take accountability? Similarly, it was mentioned that those vendors making products from collected data, should ensure they contribute back to the society it was explicitly obtained from i.e. Robot Tax.
Another way to experience A.I was through sandboxing. The sandbox approach will let you take a scientific approach allowing you to test A.I hypothesis, it will let you observe results and then let you adjust where you need to add natural learning-intelligence to help achieve your goals. Discussions around different industry use cases were present through out the talk, where taking a scientific approach to A.I programming is more likely going to let society make advances faster, such as healthcare or biometric authentication. Similarly for other industries, like driverless cars, it was discussed that it could be upto 5 years before we could trust the actual decisions any A.I system could make, where, chances are by then, an increase in cognitive systems to achieve activities will be seen.
So whilst the current the global phenomena of A.I is being explored, its actual deliverable, product, artefact or even end-goal is still being researched, in the meantime, anyone is able to access A.I data available to create stronger functionality of their field devices and products.
List of panel speakers for this event can be found on The Royal Society website here.
The inclusion of philosophy, ethics, moral responsibility into any type of computer code was discussed, as perhaps impractical, simply because a computer is not morally culpable, making machine to machine interactions biased based of the type of A.I used. So the idea that one system can have all the answers without human involvement makes A.I artificial. The need to clean biased human involvement within data sets was also mentioned. The polemic cases for A.I discussed, as was the singularity theory, where understanding the moral and ethical obligation by the system, manufacturer and user can help decide where actual responsible accountability might exist e.g. if an exoskeleton was used instead of a wheelchair, and the user experienced malfunction – who would take accountability? Similarly, it was mentioned that those vendors making products from collected data, should ensure they contribute back to the society it was explicitly obtained from i.e. Robot Tax.
Another way to experience A.I was through sandboxing. The sandbox approach will let you take a scientific approach allowing you to test A.I hypothesis, it will let you observe results and then let you adjust where you need to add natural learning-intelligence to help achieve your goals. Discussions around different industry use cases were present through out the talk, where taking a scientific approach to A.I programming is more likely going to let society make advances faster, such as healthcare or biometric authentication. Similarly for other industries, like driverless cars, it was discussed that it could be upto 5 years before we could trust the actual decisions any A.I system could make, where, chances are by then, an increase in cognitive systems to achieve activities will be seen.
So whilst the current the global phenomena of A.I is being explored, its actual deliverable, product, artefact or even end-goal is still being researched, in the meantime, anyone is able to access A.I data available to create stronger functionality of their field devices and products.
List of panel speakers for this event can be found on The Royal Society website here.