Nick Bostrom: What Happens When Our Computers Get Smarter Than We Are?
In a presentation about artificial superintelligence, Nick Bostrom speaks of a future where machine intelligence is the last invention of humans because machines will be better at inventing than humans. According to Bostrom, intelligence is merely an optimization process and machine superintelligence is a maximum optimization process. To prevent this superintelligence from removing us humans from the equation as a response to its coded protocol, Bostrom challenges that we ensure it shares our values. We can leverage its intelligence, as Bostrom explains, so its always on our side. Does the programming of human values into a superintelligence automatically assume an indication of trust as well? Do these go hand in hand. Bostrom’s presentation reminded me of classic con men. Con men are sly, confident, and manipulative. There is a difference between human values and human traits, but wouldn’t both be necessary to build trust? Even without anthropomorphizing, machine intelligence (at least in this current moment) doesn’t feel like one of us. In addition to building a superintelligence that shares our values, shouldn’t we also be cautious about letting our guards down and fully accepting it?
Murray Shanahan, “Heaven or Hell,” The Technological Singularity
In a wonderful chapter about transhumanism, AI, and superintelligence, Murray Shanahan debates the future of the singularity. He discusses a variety of topics from human and machine intelligent rights, responsibilities, and broad consequences of creating and accepting machine superintelligence into our world. He discusses scenarios where AI is at human level and progressively skilled at enhancing itself through feedback cycles of self-improvement. This made me think of today and our dependence on Google. We look to Siri and Alexa for many of our day-to-day answers – how to solve things, how to get places, how to cook things, etc. We are seeking self-improvement by inquiring into these iOT engines, so are the Siris and Alexas of the world actually machine superintelligence in the making? Are we unknowingly contributing to the expansive of a connected network of computer minds and feeding it information for it to enhance its cognition? What if the feedback loop starts with us as a building block and the machine intelligence is learning how to be a better human from all of our searches and inquiries? If a machine superintelligence evolves from what I know about the internet, I would definitely be wary.