Post Reply 
Convert Foster Netowrk to Cauer
06-06-2024, 08:07 AM (This post was last modified: 06-06-2024 03:57 PM by carey.)
Post: #11
RE: Convert Foster Netowrk to Cauer
It seems that my post in which I presented some AI-generated PPL code to help answer the op's question whether the Euclidean Division algorithm can be done on the HP Prime, posted only after noticing no other code had been posted (and included the caveat "For whatever it's worth"), attracted several off-topic AI comments. While it's not surprising that in a "Museum" some folks might look askance at the innovation of AI-generated code, and nothing wrong with that, but some comments seemed to ignore basic facts about this AI-generated code, hence this reply. Interestingly, the AI-generated code remains the only PPL code posted to this thread (on the fascinating network theorems of Foster and Cauer), took only milliseconds to generate, is easy-to-understand, and worked correctly the first time, already a significant improvement over the quality of ChatGPT generated code just a few months ago.

(06-04-2024 11:26 PM)BruceH Wrote:  
(06-04-2024 09:08 PM)carey Wrote:  For whatever it's worth, here's ChatGPT's response...

I hereby declare GodwAIn's law: as any discussion grows longer the probability of a response created using AI approaches 1.

While I'm sure it was an attempt at humor, this play-on-words that confuses including AI-generated code within a human response with an AI-generated response and analogizes to Godwin's law (a "law" that predicts the inevitability that discussions, if continued long enough, will make reference to nazis and Hitler) seems a bit much. (And if heading in that direction, missed the opportunity to mention the interesting but tragic real-life biography of William Cauer that ended in Berlin in 1945).

(06-05-2024 08:12 AM)hp-zl Wrote:  Are you serious about this artificial intelligence or are you just joking?
After all, the above-mentioned the algorithm created by artificial intelligence is pathetic.
It is division with the remainder obtained by laborious subtraction of the divisor (sic!).
If you don't believe it, give him 1000000 and 3 - it will take a while.

But back to the topic, enough of this mockery.

The challenge to try Dividend = 1000000 and Divisor = 3, (presumably intended as a torture test of this code), gives the expected result of Quotient = 333333 and Remainder = 1 in 818 milliseconds using the Windows HP Prime emulator. (Timing was accomplished by asking ChatGPT to generate additional program execution timing code. As with the original code, the timing code worked correctly on the 1st attempt and has now been added to the code in my initial post).

Criticism on the basis of inefficiency of this AI-generated code seems misplaced and ignores: (i) speed-optimized code wasn't prompted for; (ii) the code is easy-to-understand; (iii) the time to generate the code (milliseconds) was of the same order (milliseconds) as code execution time (let that sink in); (iv) for many applications, programmer time is at least as important a consideration as program execution speed.

Imagine an IT manager walking over to a programmer to task her to write a program and she hands back a working prototype before the manager finishes the sentence. It's interesting that a human programmer would be praised as a prodigy for such an accomplishment but AI-generated code is considered "pathetic."

(06-05-2024 08:12 AM)hp-zl Wrote:  PS.
I have always said that with this artificial intelligence it is "much ado about nothing".

"There is no reason anyone would want a computer in their home."
- Ken Olsen, founder of Digital Equipment Corporation (DEC), World Future Society convention, Boston, MA (1977).

(06-05-2024 11:41 AM)roadrunner Wrote:  The CAS function iquorem returns a similar result as that chat GPT program, but way faster.

You realize you're comparing the speed of a CAS function that might have taken man-hours or man-days to write vs an AI-generated function that took milliseconds to generate and wasn't optimized for speed?

======================

Finally, having never attempted PPL programming before, the experience of manually merging the timing code with the original code (both AI-generated) forced me to examine the structure of PPL code and learn a little PPL syntax, enough that I'm now motivated to learn PPL (with an AI-assistant :). In short, there's nothing to fear (this isn't skynet) and one can use AI as little or as much as one chooses, or not at all.
Find all posts by this user
Quote this message in a reply
Post Reply 


Messages In This Thread
Convert Foster Netowrk to Cauer - KeithB - 06-03-2024, 01:48 PM
RE: Convert Foster Netowrk to Cauer - carey - 06-06-2024 08:07 AM
RE: Convert Foster Netowrk to Cauer - jte - 06-07-2024, 03:09 AM



User(s) browsing this thread: 2 Guest(s)