Proceedings PDF presented July 17, 2016 at the 2016 Annual International Conference on Biologically Inspired Cognitive Architectures (New York City, USA).
Abstract: Arguably, the most important questions about machine intelligences revolve around how they will decide what actions to take. If they decide to take actions which are deliberately, or even incidentally, harmful to humanity, then they would likely become an existential risk. If they were naturally inclined, or could be convinced, to help humanity, then it would likely lead to a much brighter future than would otherwise be the case. This is a true fork in the road towards humanity’s future and we must ensure that we engineer a safe solution to this most critical of issues.
Presentation Powerpoint * Proceedings PDF presented November 9, 2015 at the 2015 Annual International Conference on Biologically Inspired Cognitive Architectures (Lyon, France).
Presentation Alternate Title: Why Your Google Car Should (Sometimes) Kill You
Recent months have seen dire warnings from Stephen Hawking, Elon Musk and others regarding the dangers that highly intelligent machines could pose to humanity. Fortunately, even the most pessimistic agree that the majority of danger is likely averted if AI were “provably aligned” with human values. Problematical, however, are proposals for pure research projects entirely unlikely to be completed before their own predictions for the expected appearance of super-intelligence . Instead, with knowledge already possessed, we propose engineering a reasonably tractable and enforceable system of ethics compatible with current human ethical sensibilities without unnecessary intractable claims, requirements and research projects.
Published Sept. 5, 2014 in Machine Medical Ethics (Intelligent Systems, Control and Automation: Science and Engineering) edited by Simon Peter van Rysewyk & Matthijs Pontier
PDF * Product Flyer * Purchase from Amazon
Implementation of ethics in medical machinery is, necessarily, as machine-dependent as ethics is context-dependent. Fortunately, as with ethics, there are broad implementation guidelines that, if followed, can keep one out of trouble. In particular, ensuring correct codification and documentation of the processes and procedures by which each decision is reached is likely, in the longer view, even more important than the individual decisions themselves. All ethical machines must not only have ethical decision-making rules but also methods to collect data, information and knowledge to feed to those rules; codified methods to determine the source, quality and accuracy of that input; trustworthy methods to recognize anomalous conditions requiring expert human intervention and simple methods to get all of this into the necessary hands in a timely fashion. The key to successful implementation of ethics is determining how best to fulfill these requirements within the limitations of the specific machine.