You can read and interpret these words because of electrical brain activity triggered by your eyes. You’re responding to sensory input in order to create meaning from this interaction with my words. I mention this because IBM and DARPA (the Defense Advanced Research Projects Agency) have created neuromorphic silicon chips, or silicon neurons, that do something similar by mimicking activity in the cortex of the human brain.
In other words, the birth of “artificial intelligence,” or AI, has arrived in concrete terms. Through the process of spike timing, these neuromorphic chips (we’ll take the liberty of calling them “neurons” as a fair analogy) become more responsive to other neurons that match their own signaling behavior while ignoring those neurons that fail to work constructively to strengthen connections. Thus, similar behavior is strengthened, different behavior weakens, and, consequently, the neurons change shape and behavior. This means neuromorphic chips can now learn through experience, or think. Everyone paying attention to this research knows that this is just the beginning of what will become an AI reckoning that will transform the planet dramatically, and much sooner than most people realize.
We should be asking ourselves all sorts of questions concerning what will happen next. For instance, DARPA’s involvement ensures military implementation of neuromorphic intelligence, both defensively and offensively. It isn’t a stretch to assume that the NSA, CIA, and FBI will be involved in a multitude of AI applications. It’s also easy to imagine a new Cold War stemming from AI innovation. But some are more concerned with what AI will do once it realizes we can’t control it. Science continues to direct humanity’s destiny in mostly positive ways, but this trajectory must be assisted by ethical guidance.
I don’t think anyone will be able to understand advanced AI behavior for a few key reasons, but mainly because of one inherent difference–we’re a carbon-based species, and AI is silicon based. No matter how hard we try to program AI in our image in order to control it, we’ll be inventing a duplication that will perceive things differently because of its inherent otherness. By default, no one can accurately predict the perceptions of silicon consciousness. What we can predict is that our neuromorphic offspring will behave in unexpected ways, like a Humboldt squid sizing up a scuba diver while deciding when to strike. I don’t really see any way of getting around this.
