Ted (merovingian) wrote,


There are many things that modern artificial intelligence can't do well. Creative leaps. Assuming things from context. Judging what's useful by looking at an immediate situation. Making broad strategies and sticking to them. Arimaa. Processing visual information.

One thing that modern AIs do very well is learning patterns from a body of data. Make a neural network with multiple layers, and program it to learn. Then give it a bunch of positive and negative samples. It will do great.

For example, you can give it a bunch of rhythmic beats that are "funky", and a bunch that are not. Train it on that data, and the neural network can tell you whether a new beat is "funky" or not. It can't tell you why, and if you look at its algorithm the whole thing won't make any sense, but it can tell you what is and is not "funky."

So here's what we're doing. First, train one neural network to recognize if a group of two to six words is a meaningful sentence in the English language. Then, train another neutral network to recognize if a sentence is "catchy" and has "that ring of truth." Generate trillions of random assortments of words, and feed them through both filters. From that, generate a dozen or so sentences that have that "ring of truth" to them.

Take those sentences and make them into a coherent system of philosophical belief. Presumably it will be a profound and moving system of thought.

...which sounds great. My job now, though, is to find some deep philosophical truth in these catchy maxims.

Here's an example: "Density is Destiny."

What does that even mean?
  • Post a new comment


    default userpic

    Your reply will be screened

    Your IP address will be recorded 

    When you submit the form an invisible reCAPTCHA check will be performed.
    You must follow the Privacy Policy and Google Terms of use.