Julia & team: I've been diving deeply into using Gemini for code generation, help with very localized code fixes, diagnostic and suggestions and the quality is very good.
The thing to keep in mind is that when using it for code generation or repair, that you Must have a feedback loop. Like it generates some code, it won't compile, feed back the error messages. Then once it compiles, create some tests for the code, feed back any discrepancies. So my end result is at the same level of quality as my own code, and sometimes better. The feedback cycle is very important for getting a good result. (Programming professionally since 1965, ham license since 1962) w8lvn On Sat, Sep 27, 2025 at 9:51 PM Julia Tuttle <[email protected]> wrote: > I'm afraid that anyone seeing huge productivity boosts in programming is > likely taking a corresponding hit in quality, whether they're noticing it > or not. LLMs make mistakes just like human coders (after all, their > training data came from humans!), and sure, they can be asked to test their > code, but they've also been caught generating bogus tests that don't > actually test the code, or using workarounds to avoid the tests being run > at all -- so you're stuck 'proofreading' the LLM's code anyway, negating a > lot of the speed benefits. > > I'm a professional programmer, and the one place I have found them > *somewhat* helpful is as (as I frequently refer to them) glorified > autocomplete. I don't trust an LLM to write larger bodies of code, and I > frankly express my programming ideas better as code than a request to a > chatbot anyway. I *do* trust them to save me some wrist strain while coding > by anticipating the rest of a line or the next few lines. I do have to read > it carefully, so it doesn't make me *faster*, but it does save some wear > and tear on my body in the process. > > 73, > > Julie > > On Sat, Sep 27, 2025 at 9:51 PM Bill Lederer via Elecraft < > [email protected]> wrote: > >> I tried that query into gemini and it gave me the Bill Winde QSO party. >> >> Regarding the utility of the LLM services, it is making a significant >> multiplier in productivity in programming. >> >> On Sat, Sep 27, 2025 at 7:28 PM Al Lorona via Elecraft < >> [email protected]> wrote: >> >> > For two evenings I've been hearing folks calling "CQ BW" in what is >> > obviously some kind of contest. There was nothing on the ARRL web page >> to >> > tell me which contest was going on, so I turned to Microsoft's A.I., >> > Copilot. >> > >> > I submitted the following prompt: "What HF contest is going on right >> now? >> > I hear people calling 'CQ BW' on 40 meters." >> > >> > Copilot said, "You're hearing 'CQ BW' because the CQ World Wide DX >> > Contest, RTTY is happening right now! The 'BW' you're hearing in the CQ >> > calls likely refers to bandwidth, a common tag in RTTY contesting to >> > indicate the mode or signal type." >> > >> > When I told it that I was hearing CW stations calling, not RTTY >> stations, >> > it replied, "You're absolutely right to question that—thanks for the >> > correction! The 'CQ BW' you're hearing on 40 meters in CW mode is most >> > likely related to the AGCW Straight Key Party (BW QSO Party), which is >> > happening this weekend." >> > >> > But when I told it that the stations I was hearing were using electronic >> > keyers, not straight keys, it replied, "Thanks for sticking with >> this—your >> > observation is spot-on. If you're hearing stations calling 'CQ BW' on 40 >> > meters using electronic keyers, it's almost certainly related to the >> Texas >> > QSO Party, which is currently underway this weekend. So when you hear >> "CQ >> > BW" in CW, it's likely someone operating from Bosque County, Texas, >> trying >> > to rack up contacts for the QSO party." >> > >> > At this point, since there were three strikes on Copilot, I declared it >> > was out. >> > >> > Backing up a few weeks, I had A. I. look up an equation for calculating >> > the noise floor of a system. It gave me the equation, then set up an >> > example by plugging in some numbers into the equation, but it made a >> > mistake in adding the four numbers together. When I pointed out the >> error, >> > it said, "Oh, you're right! I won't make that mistake again." >> > >> > I keep hearing how 'A.I. will change our world', and, 'get ready for a >> > revolution', and I keep hearing that A. I. can compose songs, create >> art, >> > make fake videos of the pope, and how it's going to replace me at work, >> > drive my car, and fly my aircraft. But all I've experienced is >> conjecture, >> > guessing, and outright error. I'm not impressed so far: sorry, A.I. is >> not >> > even close to being ready for prime time. >> > >> > R, >> > >> > Al W6LX >> > >> > ______________________________________________________________ >> > Elecraft mailing list >> > Home: http://mailman.qth.net/mailman/listinfo/elecraft >> > Help: http://mailman.qth.net/mmfaq.htm >> > Post: mailto:[email protected] >> > >> > This list hosted by: http://www.qsl.net >> > Please help support this email list: http://www.qsl.net/donate.html >> > Message delivered to [email protected] >> >> >> >> -- >> --w8lvn-- >> ______________________________________________________________ >> Elecraft mailing list >> Home: http://mailman.qth.net/mailman/listinfo/elecraft >> Help: http://mailman.qth.net/mmfaq.htm >> Post: mailto:[email protected] >> >> This list hosted by: http://www.qsl.net >> Please help support this email list: http://www.qsl.net/donate.html >> Message delivered to [email protected] > > -- --w8lvn-- ______________________________________________________________ Elecraft mailing list Home: http://mailman.qth.net/mailman/listinfo/elecraft Help: http://mailman.qth.net/mmfaq.htm Post: mailto:[email protected] This list hosted by: http://www.qsl.net Please help support this email list: http://www.qsl.net/donate.html Message delivered to [email protected]
