While I have tried various LLMs for code (such as C code to create a TCP server socket, and to accept a connection) I'd argue it is not generating to any significant degree--searching hard enough, I found the original code I'm pretty sure the LLM plagiarised.
LLMs do not understand (by my definition of the word). And the old trick of asking "how many Ws are there in Mississippi" would not be so no popular if they did understand. Roops --- "Mundus sine Caesaribus" On Fri, 29 Aug 2025, 02:35 Seymour J Metz, <[email protected]> wrote: > I believe that current technology is good enough to generate decent HLASM > with adequate training. > > As to the future, something that I learned a long time ago is that it is a > lot easier to predict that a technology is possible than it is to predict > when. Some things arrived a lot sooner than anybody expected, while others > that were "5 years away" still aren't here after half a century. > > As for true AI, I'm more concerned about ethical issues than I am about > competition. At what point does a piece of software become a person? > > -- > Shmuel (Seymour J.) Metz > http://mason.gmu.edu/~smetz3 > עַם יִשְׂרָאֵל חַי > נֵ֣צַח יִשְׂרָאֵ֔ל לֹ֥א יְשַׁקֵּ֖ר > > > > > ________________________________________ > From: IBM Mainframe Discussion List <[email protected]> on behalf > of Steve Thompson <[email protected]> > Sent: Thursday, August 28, 2025 9:07 PM > To: [email protected] <[email protected]> > Subject: Re: [EXTERNAL] Re: Artificial stupidity in action > > > External Message: Use Caution > > > I'm not so sure it has the smarts today to generate assembly > language statements that are correct and logical until we go > through another generation or two (Versions in product speak). > > I've been looking at it in reading IRS alc and trying to turn > that in to Java. I was told the java was nearly undecipherable by > the people who were well versed in java. And even then, they were > not able to make some changes to it that did what they wanted. > > Meanwhile, I'm afraid of AI becoming too "good". I keep > remembering the books I read in HS from Isaac Asimov and > postitronic brains and having to implement the rules of robotics. > > Basically, I see a bad moon rising.... > > > -- > Regards, > Steve Thompson > > MAGA: Make Assembly Language Great Again > > > > On 8/28/2025 8:44 PM, Seymour J Metz wrote: > > Just because it's overhyped doesn't mean that it's not useful. And just > because it doesn't have understanding today doesn't mean that it never > will. Making a pun on the name doesn't mean that I consider it useless or > unimportant. > > > > The issues generating HLASM code are real, but that stems from > inadequate data during training rather than design defects. > > > > -- > > Shmuel (Seymour J.) Metz > > http://mason.gmu.edu/~smetz3 > > עַם יִשְׂרָאֵל חַי > > נֵ֣צַח יִשְׂרָאֵ֔ל לֹ֥א יְשַׁקֵּ֖ר > > > > > > > > > > ________________________________________ > > From: IBM Mainframe Discussion List <[email protected]> on > behalf of [email protected] < > [email protected]> > > Sent: Thursday, August 28, 2025 8:09 PM > > To: [email protected] <[email protected]> > > Subject: Re: [EXTERNAL] Re: Artificial stupidity in action > > > > > > External Message: Use Caution > > > > > > The same people who said the mainframe is dead in 1993, are likely the > same ones now downplaying AI. NVIDIA is the highest market cap company that > ever existed. What do they make? Graphics Processing Units for AI data > centers. Who is buying them? The largest tech companies in the world. AI is > here and it’s going to impact everything. > > > > > > Sent from Yahoo Mail for iPhone > > > > > > On Sunday, August 17, 2025, 8:59 AM, Seymour J Metz <[email protected]> > wrote: > > > > How good is the training? Can it generate a working HLASM program? > Locate specific editions of PoOps? > > > > i tried CoPilot only because it was part of my mandatory training this > year. i was curious whether it would mess up as badly as ChaatGPT (It's a > typo. That's my story and I'm sticking to it.) > > > > -- > > Shmuel (Seymour J.) Metz > > http://mason.gmu.edu/~smetz3 > > עַם יִשְׂרָאֵל חַי > > נֵ֣צַח יִשְׂרָאֵ֔ל לֹ֥א יְשַׁקֵּ֖ר > > > > > > > > > > ________________________________________ > > From: IBM Mainframe Discussion List <[email protected]> on > behalf of Pommier, Rex <[email protected]> > > Sent: Saturday, August 16, 2025 10:49 PM > > To: [email protected] <[email protected]> > > Subject: Re: [EXTERNAL] Re: Artificial stupidity in action > > > > > > External Message: Use Caution > > > > > > How about the best of both worlds - google AI? I went searching for a > very specific "can I do this" type question regarding surcharges on retail > purchases with credit cards and a specific CC clearinghouse. Google AI > came back with a "yes you can do that" and about 3 items down the hit list, > the specific clearing house' web site pops up with "no you can't do that". > > > > Rex > > > > -----Original Message----- > > From: IBM Mainframe Discussion List <[email protected]> On > Behalf Of Matt Hogstrom > > Sent: Friday, August 15, 2025 4:38 PM > > To: [email protected] > > Subject: [EXTERNAL] Re: Artificial stupidity in action > > > > Let’s be honest. It’s not sentient. Processed a lot of information and > indexes it to define relationships and is a better search than google. > > Apart from that I think people are expecting perfection but it is > imperfect but pretty dang impressive > > > > On Fri, Aug 15, 2025 at 17:34 Seymour J Metz <[email protected]> wrote: > > > >> Well, the judge thought that submitting briefs with bogus citations > >> was pretty cheesy. > >> > >> -- > >> Shmuel (Seymour J.) Metz > >> https://urldefense.com/v3/__http://mason.gmu.edu/*smetz3__;fg!!KjMRP1I > >> xj6eLE0Fj!o_VDCnJjynb2bLfg3a11_uUWQ7D6fOm5XPAHuwxjVSyRa1TXm8EZK0TBPkbq > >> Iq9QkXgdysUR5N9aWYQ8nUZqrh1AdEOYmjoWbmAO$ > >> עַם יִשְׂרָאֵל חַי > >> נֵ֣צַח יִשְׂרָאֵ֔ל לֹ֥א יְשַׁקֵּ֖ר > >> > >> > >> > >> > >> ________________________________________ > >> From: IBM Mainframe Discussion List <[email protected]> on > >> behalf of Jeremy Nicoll <[email protected]> > >> Sent: Friday, August 15, 2025 4:58 PM > >> To: [email protected] <[email protected]> > >> Subject: Re: Artificial stupidity in action > >> > >> > >> External Message: Use Caution > >> > >> > >> On Fri, 15 Aug 2025, at 19:42, Seymour J Metz wrote: > >>> ... when they submitted AI-generated brie that contained > >> AI can make cheese? > >> > >> -- > >> Jeremy Nicoll - my opinions are my own. > >> > >> ---------------------------------------------------------------------- > >> For IBM-MAIN subscribe / signoff / archive access instructions, send > >> email to [email protected] with the message: INFO IBM-MAIN > >> > >> > >> > >> > >> ---------------------------------------------------------------------- > >> For IBM-MAIN subscribe / signoff / archive access instructions, send > >> email to [email protected] with the message: INFO IBM-MAIN > >> > > ---------------------------------------------------------------------- > > For IBM-MAIN subscribe / signoff / archive access instructions, send > email to [email protected] with the message: INFO IBM-MAIN > > > > ---------------------------------------------------------------------- > > For IBM-MAIN subscribe / signoff / archive access instructions, > > send email to [email protected] with the message: INFO IBM-MAIN > > > > > > > > ---------------------------------------------------------------------- > > For IBM-MAIN subscribe / signoff / archive access instructions, > > send email to [email protected] with the message: INFO IBM-MAIN > > > > > > > > > > ---------------------------------------------------------------------- > > For IBM-MAIN subscribe / signoff / archive access instructions, > > send email to [email protected] with the message: INFO IBM-MAIN > > > > > > > > ---------------------------------------------------------------------- > > For IBM-MAIN subscribe / signoff / archive access instructions, > > send email to [email protected] with the message: INFO IBM-MAIN > > ---------------------------------------------------------------------- > For IBM-MAIN subscribe / signoff / archive access instructions, > send email to [email protected] with the message: INFO IBM-MAIN > > > > ---------------------------------------------------------------------- > For IBM-MAIN subscribe / signoff / archive access instructions, > send email to [email protected] with the message: INFO IBM-MAIN > ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN
