On Wed, 21 Aug 2024 05:41:40 -0400, zen cycle
Post by zen cycleI will admit, if I had the time I Linux is something I've thought I
might enjoy tinkering with. Maybe when I retire. I have way too much
going on in my life right now to take on another hobby.
I've been involved in debates that revolve around "What will it take
to make Linux a mainstream operating system"? The Linux fanatics will
propose major changes to the licensing, UI, standards, etc. When I
suggest that all it will take is for Microsoft and Apple to make a few
big mistakes, I'm usually greeted with silent contemplation.
Post by zen cycleWe had an intern a short while ago who was tasked with writing code for
a somewhat simple of test rig we were using for development. It required
a rather large look-up table with several input parameters. He finished
it in about ten minutes, and it worked. I said "wow, that was quick". He
replied, "I used chat GPT".
Be that as it may, this kid _was_ (_is_ actually) quite bright -
enthusiastic, inquisitive, and resourceful. I really couldn't fault him
for using a tool to create the large table (hundreds of elements) rather
than take the better part of a day to create the whole thing by hand. It
would be one thing if he didn't understand what the AI generated, but he
was insightful enough to point out a tweak to the algorithm in the code
of the product under development would give better performance after
looking at the output data from the test jig. The lead software engineer
on the project agreed with him.
Very good. He'll do well. However, there might be a problem. Long
ago, I did the same thing. I was working on a radio direction finder
(AN/SRD-22). I contrived a way to greatly simplify the active antenna
and associated driver circuitry. I threw together a prototype and
demonstrated it to everyone who would listen. Just one problem. The
project was well past the design stage and we were soliciting quotes
of components. If we were going to use my scheme, it would bring
development to a screeching halt and risk missing the delivery
deadline. Management held several high stress meetings (to which I
was NOT invited) and eventually agreed that my redesign was best. The
result was everyone in engineering had to put in substantial amounts
of overtime to meet the deadline. Everyone in engineering wanted to
kill me, but obviously, that didn't happen. Anyway, beware of good
ideas and bad timing.
I ran into something similar when CAD was introduced as "design
automation". Management somehow decided that buying CAD workstations
would allow them to hire minimally educated people to take the place
of higher paid designers. I know of one company that took the plunge.
They jumped into the CAD swamp running. They laid off their high
priced design staff, purchased some version 1.0 CAD equipment, and
hired some enthusiastic the clueless computer operators. It took a
few years, but eventually, it killed the company:
<https://www.latimes.com/archives/la-xpm-1987-10-15-fi-14057-story.html>
The moral was that you can give a CAD system to a good designer and
they become a better designer. However, giving a CAD system to a
beginner doesn't automatically make them a product or PCB designer.
The same thing applies to AI. The 16 year old in my previous rant is
very smart. He taught himself several computer languages. I taught
him a few things about PC hardware. He ran with that and now knows
more about recent computer hardware issues and problems than I do. His
use of AI for programming is an extension of what he already knows
about programming, not a replacement. I expect him to do very well.
Meanwhile, some of the older technical types, that hang around FLUG
(Felton Linux Users Group) and the local radio clubs consider AI a
threat to themselves and the world.
--
Jeff Liebermann ***@cruzio.com
PO Box 272 http://www.LearnByDestroying.com
Ben Lomond CA 95005-0272
Skype: JeffLiebermann AE6KS 831-336-2558