Elemental AI 11/15/23
----------------------------------------------------------------------
I'm no expert on artificial intelligence. Which is another way of
saying, I am not qualified to write on this subject but I'm going to
anyway. To make matters even worse, I'm not well read on this subject,
so you'll be getting nothing but opinion here.
For people like me, the term AI has long implied actual independent
intelligence, a complex ability to make decisions without instruction,
otherwise known as free will. Lately, it seems to me, the term has
been used to describe the ability to trick humans into thinking
programs have this ability; or, to describe complex programs that
perform impressive parlor tricks that appear human-like (by using
massive amounts of human-generated inputs).
If you accept the AI description as it is attached to things like
ChatGPT, it seems to me (the non-expert) that you're lowering the bar
of human achievement from creating artificial intelligence, or free
will, to simply tricking people into thinking that you've done so. Not
to minimize the accomplishment, which is immense... but do we have to
accept it as something it is not?
That got me to thinking about all the things we can now call AI. Take
typesetting as an example. Previously, an intelligent human had to
determine all kinds of things, such as when to break a line. Now, AI
can instantly determine where to break a line as I resize my browser
window, or any other window. It's intelligent! And that's just the
beginning of this sort of elemental AI.
Ultimately, the kind of AI that is ChatGPT isn't much more impressive,
to me, than this ability for a computer to decide to break lines. Yes,
I realize that a lot more work and data goes into ChatGPT. But, it's
no closer to being independent will than the routines that break my
lines for me. It's no more impressive in that sense.