[gurps] Failing autopilots and RVO

[Knapp]

On Wed, Mar 4, 2009 at 3:09 PM, Bira <u.alberton at gmail.com> wrote:
> On Tue, Mar 3, 2009 at 7:20 PM, Travis Watkins <terwin3 at gmail.com> wrote:
>> I would not count on that.
>>
>> There are already specialized programs that are better than humans in
>> a number of fields, including medical diagnosis(saw an article on this
>> some years ago, but people still want a human, so some doctors use it
>> as a second opinion/sanity check)
>
> That's a rule-based expert system, IIRC. It can make decisions based
> on its pre-programmed rules and on properly-formatted input, yes, but
> that's all it does. It can't really be considered "intelligent" in the
> same way a person can, which is what I assume is the "real thing"
> that's being referred to here.
>
> --
> Bira

Funny how doctors get payed so much for not being intelligent but just
being experts systems. I am not being sarcastic. We seem to get payed
more when the task does not take skills that are easy for a human like
memorizing decision trees. It is called differential diagnosis and
really is quite hard to do well. I am currently studying medicine so I
can say this with real knowledge.

I also read about that expert system in a Prolog programming book. The
programmers were upset because their systems success rate was so low,
so they decided to test real doctors to see what they should be
shooting for. The doctors did much worse than the expert system. I
think I bought that book in 1986. I wonder how much better the systems
are now?

-- 
Douglas E Knapp

Why do we live?