Can we ever know the truth? If so, how? These are ancient questions. Unanswered for millenniums, we have now impatiently added another one: "If we ever figure out the solutions for those two, are we smart enough to go one step further and build a system capable of knowing the truth, i.e. ‘artificial intelligence’?"

Let’s trace back our endeavor thus far. Rene Descartes started by questioning everything, but he still found at least one thing he could be sure of: "I think, therefore I am."¹ Not a bad start. At least, I exist. John Locke insisted that our mind at birth was just a blank slate (“tabula rasa”), only filled later through experience². Even for this, inadvertently, he had to assume that we are born with some kind of method to fill our mind with experience. So far so good. Not only I exist, now I can also experience the world. David Hume, known for his skepticism, forcefully argued that we could never rationally justify any conclusions beyond our past experience ("the problem of induction")³. As bleak as this may sound, I choose to look at the silver linings. Firstly, I don’t think he had any real quarrel with either Descartes or Locke. Rather, his concern was that experience was ALL that we could assuredly have. Secondly, by this seemingly hopeless conclusion, he had unwittingly pointed us to the final piece of the puzzle: "a little something", which we can be absolutely sure of, that will take us beyond our direct experience.

It has taken us another 250 years. I’m having a strong suspicion that we have finally advanced enough to the tipping point that if we can just add this "little something," things will start clicking. I don’t claim to be worthy of being mentioned in the same breath as those three, but I am here to say that this "little something" is "coherence".

What is coherence? Simply put, coherence is internal consistency and external relevancy. There is nothing magic about it. It’s basically making sense. We all do that ourselves and demand of each other. As a matter of fact, we can’t help but do that (even though my sense may or may not be the same as yours). Not only that, whenever you have any doubts, you are demanding that things make sense to you, i.e. being coherent. So, not only you can’t doubt the doubter (à la Descartes), neither can you doubt the doubting, which both requires and demands coherence.

What do we know about the unknown? There are two kinds: the knowable unknown and the unknowable unknown. In our pursuit of truth, we are only interested in the former because there is nothing we can do about the latter. What are we sure about the former? Since they are knowable, there is something to be known about them; there is some sense to be made about them. In other words, they are sensible or intelligible. Or simply, they are coherent, and they are not nonsense. Let’s do a little thought experiment. Say we are beamed to a totally strange place and meet aliens there whom we didn’t know anything about, what is the only thing we can hope to rely on to communicate with them and try to figure out their intentions, and vice versa? That they are coherent! Because, by asking this very question and making such an attempt, we have already assumed that they have intentions and the intentions are knowable to us. If they are not? Oh well, there isn’t anything we can or should do. They are not that interesting, more like rocks than us.

I am sure most of you are not convinced yet. That’s good, because I am not here to convince you. I happened to pick up a clue from 250 years ago and then tried to set up a bar. If we create a system that can surpass that bar and it then proceeds to find some truth that we have been seeking but were never successful ("the moment of truth?"), it speaks for itself. If it looks like a duck, walks like a duck and talks like a duck, all the time and everywhere, it has to be a duck. That’s what a duck IS. Stay tuned…