From: blockedofcourse@foo.invalid   
      
   On 2/2/2026 11:47 AM, bitrex wrote:   
   > It's pretty useful for debugging software, I can feed it a few hundred lines   
   of   
   > code and it can find the missing brace that's causing a cryptic C error   
   message   
   > much faster than I can. It's also pretty good at auto-generating   
   documentation.   
   > Who likes writing documentation?!   
      
   Those are all LLM related tasks.  That's the problem with the "chatbots";   
   they have conditioned everyone into thinking that's (all) AI is.   
      
   > It's a good academic-to-English translator. A number of textbooks and   
   > whitepapers on e.g. physics and mathematics either just aren't very   
   > well-written or make certain assumptions about the reader's background and   
   > leave out steps, and GPT is adept at filling in the blanks for the confused.   
   >   
   > Two years ago GPT really stunk at math but it's come a ways and it makes a   
   > pretty good ad-hoc computer algebra system. It's sort of remarkable that it   
   can   
   > do this without actually connecting to a "real" CAS on the back end, as I   
   > understand it the transformer algorithm/neural network somehow constructs a   
   > rudimentary CAS out of nothing on-the-fly, which is pretty remarkable when   
   you   
   > think about it. Example:   
   >   
   > "I would like a resistor divider made up of 5 resistors in series, grounded   
   at   
   > one end, and with some voltage going into the series string. Please devise   
   > constraints on the values to produce outputs at the taps where each output   
   > starting from the bottom, is approximately double the previous. Then select   
   1%   
   > standard tolerance values for a total string resistance of 100k."   
   >   
   > Output: 6.34k, 6.34k, 12.7k, 24.9k, 49.9k. Looks good to me..   
   >   
   > However it's kind of terrible for more free-form electronics design tasks and   
   > will often lead you astray about very simple circuits without close   
   supervision.   
      
   *Learning* algorithms will be the biggest boon to electronic design.   
   Making devices that anticipate your needs, handle the annoying   
   little details, etc.   
      
   We'll likely see a whole new generation of devices that don't just "do   
   what they are told" but anticipate needs and even suggest differences   
   in usage patterns.   
      
   My "garage attendant" lets me know if any of my exterior lamps on the   
   car are "broken".  In addition to warning me if something is breaking the   
   plane of the garage door (opening or closing) that the "electric eye"   
   can't see -- like the nose of the car (backing in) not far enough into   
   the garage or someone on a ladder accessing the "attic" space as the   
   door is commanded open.  Or, something IN the garage before it allows   
   the door to open (it sends video of the scene to the car via a WiFi   
   link) so you don't "let the dog escape" or discover something in   
   the path of the vehicle as you enter).   
      
   My HVAC controller kept giving me "bogus" results while I was developing   
   the learning algorithms for it.  Actually, it was telling me that the ACbrrr   
   was failing and I thought it's results were in error (the ACbrrr failed   
   within a month!)   
      
   My irrigation controller told me about a leaking valve because it   
   was able to "watch" non-stop.  (I need a better water meter to get   
   finer-grained usage information, especially for the "indoor circuit")   
      
   I suspect I could extract a fair bit of information from vehicles   
   just by accessing sensors that are already in place (but, my interests   
   don't lie, there).   
      
   Colleagues and old clients are already inquiring as to what they can   
   do and how to do it.  None of these things are "essential".  But,   
   neither is a GPS for your vehicle or  coffee maker that turns itself   
   on in anticipation of your morning brew.  IoT and leaf processing   
   will be the next decade's focus.   
      
   Today, a machine without network connectivity is a paperweight.   
   Tomorrow, a machine that doesn't learn will be similarly described.   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)