Sign up FAST! Login

Why You Should Be Eager to Be Wrong


Stashed in: Creativity, #success, Decisions, Awesome, Your argument is invalid., Success, Decision Making

To save this post, select a stash from drop-down menu or type in a new one:

True this:

"Uncertainty is a very mentally demanding, and in a certain way, physically demanding process."

Loved this article.  As F. Scott Fitzgerald said, "The test of a first-rate intelligence is the ability to hold two opposed ideas in mind at the same time and still retain the ability to function."  I feel like Bertrand Russell said something similar but can't place it right now.

One way to improve one's facility with uncertain/conflicting evidence is to study the basics of fuzzy logic.  The basic math is simple -- accessible to almost anyone with a little effort -- and beautiful.  In a nutshell, truth values can be 0 (false), 1, (true), or anywhere in between.  The middle values can represent probabilities, but than can also represent partial attainment. 

For example, am I close to your house?  We can answer that with a fuzzy value... 0 when I haven't left, 1 when I'm at your house, and increasing gradually from 0 to 1 as I approach.

The coolest thing to me is how fuzzy logic extends, rather than conflicts with, classical Boolean logic.  For example the fuzzy OR function is just the minimum of the 2 truth values.  But this same function (min) gives same results as a classical Boolean truth table (e.g. 0 OR 1 = 0, which is the minimum of 0 and 1)

Fuzzy if-thens can be very powerful, and have been the basis for relatively simple software solutions that can be very difficult to implement with classical logic.

Bart Kosko has a pop science intro book on this, which is good.  And I'm sure there are some good introductions online too.

That's a pretty fascinating model.

Here I thought machines were nothing but zeroes and ones. 

There's actually a lot more to it than that. 

So now I'm beginning to believe that AI really can make good decisions. 

Is that a bad leap?

I guess it depends on how you define AI... I don't think we're in any real danger of it soon, but this fuzzy model has been huge for problems under that heading.  I believe that it was fuzzy control systems that actually turned the tide on digital appliances for example... could be wrong about that.  Here's an example of a fuzzy control system based on fuzzy if-then rules... you can see how intuitive this would be versus "binary" thinking about when to alter various system parameters.  The model can be set up by a human and then the system can be trained to find the right "weights" (fuzzy values), without anyone ever having to work with the raw numbers. 

normalized-tmp-574dfd95b86b9

This seems like a much better model of how humans often think about things, i.e. rather than "all or nothing" (Boolean/binary) we think things to various degrees.  Maybe inside our brains, the thresholds for neuron firing *could* be quantified, but we never experience it this way... and yet, the system works even we we never know the specific quantities involved. 

It does seem like a better model than human typically use. 

Maybe you're right that this is how our internals work but they only manifest themselves through intuition. 

You May Also Like: