Re: Newbie Questions: Starting a Career in AI

Don Geddis <[EMAIL PROTECTED]> wrote in news:comp.ai on Tue, 02 Sep 2003:
>
> I wrote:
>> > Especially if you're going to look at AI history, and you want to
>> > experiment a lot, you should almost certainly spend some time with Lisp.
>> > Many, many AI systems were (and continue to be) built in Lisp.
>
> Anthony Bucci <[EMAIL PROTECTED]> writes:
>> True enough, but LISP is nearly dead.
>
> You are mistaken.  It doesn't have the fad-ish roller-coasters
> of popularity that other languages do.  But it has a solid core
> of users, developers, and vendors that continues through today.

http://mentifex.virtualentity.com/lisp.html is a Lisp AI Weblog.
>
>> Many schools have replaced it with more modern languages.
>
> This implies that Lisp is "old technology", and "better"
> languages are used today.  That is also in error.  There are
> more popular languages today, and ones with larger libraries.
> But for the design of the core language, Lisp
> remains superior to almost everything out there.
>
>> Brandeis teaches its intro AI courses in Scheme,

http://mentifex.virtualentity.com/scheme.html is a Scheme AI Weblog.

> but they are debating whether to switch to another language (like Java).

http://mentifex.virtualentity.com/java.html is the Java AI Weblog.

> I think the only thing stopping a change like that is inertia:
> many professors don't know Java and don't want to teach in it.
> The upper-level AI courses are usually in the professor's
> favorite language, which can range from ML to
> Python to Java.

http://mentifex.virtualentity.com/python.html is a Python AI Weblog.
>
> Here you are talking about popularity.  Lisp isn't especially
> popular.  But it is of very high quality, and one of the best
> languages for rapid prototyping, and for solving the most
> difficult computational problems (where it isn't clear that
> there even is a solution, much less what it might be).

http://mentifex.virtualentity.com/weblog.html -- AI has been solved.
>
> That, combined with its history in AI, means that someone interested
> in AI (like the original poster) would be well served learning Lisp.
>
>> It's clearly worth learning and understanding LISP, but I
>> personally wouldn't write anything in it.  I tend to think
>> it's something you'll pick up when you need it.  It's much
>> more important to be a quick study and be able to pick up
>> a new language rapidly.

http://mentifex.virtualentity.com/acm.html -- learn as you code AI.
>
> You seem to be implying that all languages are roughly equivalent,
> so you could always pick it up later if you need to.
>
> You're missing the possibility that there might be more advanced
> computational concepts in some languages.  That it's not simply
> a matter of learning a new way to write a familiar concept, but
> that you might actually have to understand something about
> computation that you've never encountered before.  That's part
> of the point of learning Lisp: to appreciate what a language
> could possibly offer.

http://mentifex.virtualentity.com/alife.html -- code the main AI loop.
>
> I don't mean to go so far as to say you cannot do AI without Lisp,
> nor that Lisp is the only language with good ideas.  But "Common Lisp"
> in particular is a very large language, with a high density of excellent
> ideas, and is closely tied to AI.  The original poster was interested in
> educating himself in preparation for AI.  Learning Lisp would be much
> more useful than calculus, or physics, or many of the other things
> he was considering.  And much more useful than learning C.
>
>        -- Don
> _______________________________________________________________________
> Don Geddis                http://don.geddis.org/         [EMAIL PROTECTED]
>
-- 
http://www.kurzweilai.net/mindx/profile.php?id=26 - Mind-eXchange