Hi Saif,
the issue is that AI (as used in the context of this discussion) does not KNOW the answers to your questions, however specific they may be.
ChatGPT generates answers by generating text that is likely to look/feel like an expected answer to your question.
So it CAN be right, but there is no guarantee for that.
As such, all answer should be cross-checked and verified before being posted to a forum like this, becaus people here expect to communicate with people and exchange knowledge with them rather than use this forum as a search engine or like ChatGPT (or any other AI).
If someone wants to help others here it is fine to use AI, but only if it is used to educate yourself before posting the answers here, in order to reduce fake/wrong solutions which are frustrating to work through.
Like the example Andrey provided, answer are mostly looking real and true, but that is no guaranteed.
------------------------------
Christian Schwabl
------------------------------
Original Message:
Sent: Thu December 05, 2024 11:50 PM
From: Saif Ali Sabri
Subject: Forum rules on AI generated content?
Mr. Andrey Klyachkin
While ChatGPT itself doesn't inherently limit search responses by time, you are general man for ChatGPT and I can provide targeted assistance, because I am an expert in how to use ChatGPT.
There are some specific methods to get the answer from ChatGPT within limitations. I am able to discussing ways to optimize ChatGPT for specific searches within defined time frames.
You can also request that ChatGPT limit its responses to the most recent updates or techniques, perhaps including only sources, documentation, or developments that are from recent months.
If you need insights from a specific time frame, within the past six months or two years, I can use my browsing capability to usage of ChatGPT gather the most relevant and recent information within this time frame.
If you provide me with more details on the specific area you're targeting (e.g., troubleshooting socket communication in IBM iSeries subsystems or optimizing communication), I can help guide the query to refine the results more effectively.
------------------------------
Saif Ali Sabri
Original Message:
Sent: Thu December 05, 2024 04:27 PM
From: Andrey Klyachkin
Subject: Forum rules on AI generated content?
I asked ChatGPT today to describe socket-based communication between srcmstr process and subsystems. The documentation has been freely available for years on IBM's site.
It described it and the description looked very good, but there were some problems:
- it told me that srcmstr creates sockets in /var/tmp instead of /dev/.SRC-unix
- it told me about imaginary message types that don't exist
- it told me that this is a bi-directional communication between srcmstr and subsystems. I'd like it to be, but if I read IBM's documentation correctly, subsystems only answer to srcmstr's requests
I asked ChatGPT to provide an example of a C code for socket communication with srcmstr. It generated code which again looked very plausible but was completely wrong.
Regarding AIX, I don't know any AI tools that could provide real working answers.
------------------------------
Andrey Klyachkin
https://www.power-devops.com
Original Message:
Sent: Fri November 29, 2024 09:56 AM
From: Russell Adams
Subject: Forum rules on AI generated content?
It's exhausting how much AI generated slop there is all over the Internet.
As a forum for users and experts on IBM software, I think that AI generated answers don't belong here.
The moderators might consider whether a rule is applicable.
------------------------------
========================
Russell Adams
https://adamssystems.nl/
========================
------------------------------