Do Humans Have The Capacity For Safe Artificial Intelligence (AI)? Posted on : Jan 18 - 2018

Innovation and technology advancements are replete with unintended consequences. In human myth, in fashioning the intricacies of the labyrinth, Daedalus made it almost impossible for Theseus to slay the Minotaur, the monster dwelling inside. Critical we consider here the long-term consequences of our inventions with great care, if for no other reason to be sure we do not find ourselves made captive therein. 

 The volatility of cryptocurrencies is something we hear about a great deal. However, the volatility is just so much noise. The real story is why cryptocurrencies are becoming so big. The answer, in part, is that recent experiences, such as the Great Recession, have contributed to a mistrust of existing institutions and systems. Instead, there is evidence of a greater confidence by Millennials in technology-based counterparts.  Faith in humans has been lost, so in machines we trust (See: What to Learn from Millennials' Distrust of Banks). With this shift, there is a need to be mindful of both what is being handed over and how that transfer is taking place. There is clearly risk in the shift that demands an awareness of the greater context. Relative to AI, we offer the following observations below:

 1) Ubiquitous Autonomous AI is a certainty

The dynamics of being able to control information and technology has long gone by the wayside. Open source development and the cooperative sharing of ideas across the globe via the internet render any ability to contain development ineffective. To our view, the internet-of-things (IoT) will be the mechanism of diffusion for ubiquitous autonomous AI throughout society. (For more see Zuckerberg and Musk Clash Over the Future of AI)

 2) The current AI paradigm is still heavily focused on task specific utilities, thereby comporting with the AI “training” paradigm

At some point, there will be algorithms that “think” in the sense that they will actively acquire real world information instead of being fed it. At that point, there will be a period of massive learning acceleration as big data becomes a well for accumulated experiences that can be shared and propagated. It is as if you had an experience that once learned could be shared by transplanting it to others. View More