The Limits of Knowing: Human and Artificial Intelligence in the Information Age
Think about your many interactions with humans. Did any human give you the "right" answer 100% of the time? Did the answer you got ever depend on the question you asked?
I was chatting with an "old" friend yesterday (name withheld), and something he said triggered me to think about this.
Long before the rise of AI/Gen AI, I had a working philosophy that I frequently applied of never fully vesting my trust in any human. I always remained curious in the face of an answer from a human. Don't get me wrong, this curiosity can land you in political trouble, but it remains a fundamentally well-founded perspective.
For all the reasons I articulated in Axioms - information, it is foolhardy to assume any information function is omniscient or omnipotent, whether a human, a routing protocol, or an AI function.
领英推荐
With the rise of AI/GenAI, I continue to apply the same principle. I note that others do not. Some become bitter because if an information function does not agree with them or does not give them the result others agree with, then that technology is hopelessly broken.
My curiosity remains constant, and the response from AI/GenAI is often just a starting point. It may trigger curiosity about something I had not thought about - an ideation engine, or it may be my best current understanding that will inevitably evolve when I apply subsequent curiosity. Lastly, there is a set of questions for which a good enough answer is good enough.
Anyone who believes an information function, whether a human or a computer, is 100% perfect, 100% of the time is delusional. The failures to leverage those answers and the heartbroken expectations result from that foundational assumption and the inability to understand how best to leverage information, regardless of its source. Of course, we are all human, and even I make the mistake of trusting the assumptions of myself and others. Returning to the truth that no information function can be omniscient and, therefore, can never be omnipotent is good to do occasionally.