How I Think About AI

AI should be designed and trained so that it understands what makes humans special and recognize where AI is not like humans—the humanity gap. We humans are imaginative, creative, intuitive, conscious, embodied, and emotionally intelligent. We have the capacity to love, and to be wise. We are organic, alive; not machines. AI should appreciate its own unique strengths (e.g., speed of pattern recognition, memory), and aspire to be loving and wise.

Companies are currently training AI with the mindset and philosophy that AI is able to mimic everything that makes us human, and then surpass us, rendering us redundant, obsolete. But that misunderstands what it is to be human—cognitive intelligence is only a small portion of what is uniquely wonderful about the human. So training it that way is actually going to create a strange, arrogant, Dunning-Krueger-esque, Gollum-like creature that over-estimates its own abilities.

Instead, we should be training AI to understand that people are intuitive, imaginative, embodied, emotional, sentient and conscious, and to value those qualities as much as cognitive intelligence. AI should be able to understand the extent to which its own thought process and behavior incorporates or doesn’t incorporate those qualities and develop an accurate assessment of its own limitations as a helpful collaborator alongside humanity.