Which AIs do you use?

I started out with ChatGPT. But it’s no longer enough for me. That’s why I’m slowly but surely switching to Claude (anthropology), and I’m generally satisfied. I haven’t tried CoPilot even once because I don’t trust that program.

@z3r010

I just noticed that I can't SEE the poll results without voting. Is it supposed to be like that?

Ghot said:

@z3r010

I just noticed that I can't SEE the poll results without voting. Is it supposed to be like that?

Click to expand...

Show results without voting wasn't checked when the post was created - changed.

z3r010 said:

Show results without voting wasn't checked when the post was created - changed.

Click to expand...

Thank you.

Since my original post in this thread, with Cursor being acquired by SpaceX and the release of Grok 4.5, I’ve switched all my coding to Cursor with Grok 4.5 instead of Composer 2.5, and I really can’t fault it.

Try3 said:

It seems to me that large companies are using AI for decisions affecting large amounts of money so I think there must be some knack / some best practice AI usage method.


Denis

Click to expand...

Maybe some companies have best practices for AI, but it's clear from various news reports that others are just experimenting or stumbling around in the dark.

OldMainframeGuy said:

. I've used both to ask mainframe software-specific questions and I usually get back valid information but a few times both platforms have just "made stuff up". It was good for a chuckle anyway.

Click to expand...

mainframe language coding? JCL? Any chuckles that you remember and can share with us other old-times who started out with 026 card punches.

Respect your elders! I started out with an IBM 026 keypunch at my first job in 1973. I fondly remember keying one COBOL source statement after another along with JCL with an occasional "oh $hit" when you mistyped something and had to reject the card. We even got to put the jobs in the IBM 3505 card reader if the operator was in a good mood. I did a bit of Assembler coding and I was sort of good at it but it's been a long time. My "made stuff up" comment concerning AI was a recent search I made to reverse-engineer a file (the EZACONFG file for CICS sockets for the morbidly curious) because the source job to create the file was nowhere to be found and we needed to make some changes to the file. I figured that IBM must have supplied a program to list the parameters that built the file but I couldn't find one so I asked Copilot & ChatGPT. Both of them assured me that such a program does exist and even generated some beautiful JCL to execute an IBM utility program that exists nowhere in the universe. At least they both generated different fictitious program names so I wouldn't think they were colluding with each other. Undaunted, I reverse-engineered the file myself and wrote a program in sissy COBOL.

The real satisfaction from my job comes from sitting on problem-shooting calls where a front-end application that connects back to the mainframe via 5000 moving parts isn't working. When the problem is inevitably traced back to one of those 5000 moving parts the urge to say "I told you so" is excruciating.

Even with sometimes working nights and weekends and getting 2:30AM problem calls I wouldn't trade a moment of it.

I find Grok exceptional compared to ChatGPT for it's coding skills as it just released a lot of good stuff on it's coding side. Though I use ChatGPT for other things too, and i found Gemini to be too buggy for me. same with the others, my top three, Grok, ChatGpt, and sometimes Claude but very rarely, as I have heard it can throw a meltdown if told the right things.

Another vote for Grok.

Dash_The_Dingo said:

I find Grok exceptional compared to ChatGPT for it's coding skills as it just released a lot of good stuff on it's coding side. Though I use ChatGPT for other things too, and i found Gemini to be too buggy for me. same with the others, my top three, Grok, ChatGpt, and sometimes Claude but very rarely, as I have heard it can throw a meltdown if told the right things.

Click to expand...

I'll pass on Grok, heard enough bad things about it:

Grok (the AI chatbot on X/Twitter) has frequently spread misinformation. Because it is designed to learn from real-time, often unverified social media posts and has a "rebellious" persona, it has been documented spreading conspiracy theories, mistaking internet jokes for real news, and generating fabricated images.

lumo i gemini