AI need not maximize a utility function

The Go playing AI recently developed has no utility function. Instead, it has a value network which learns what choices have utility. One could argue it is a utility function, but I can easily imagine a value network that bluffs or sandbags and pretends to choose a path of low utility or actually chooses a path of low utility to aid actualization of other values.

As a human, I have the luxury of deciding my hierarchy of values using information based on reality or any other ideology or superstition.

Advertisements

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s