Using ChatGPT to Examine COA's

jmk909er

jmk909er

Explorer
Joined
Jul 21, 2026
Messages
53
Reaction score
0
Location
Woodward, Oklahoma, USA
Janoshik's COA's tend to be fairly simple in most cases. When a COA is more involved—graphs, additional parameters—I give ChatGPT a copy of my own COA and request an analysis. The result is quite comprehensive, and it brings up questions that would not have occurred to me. That makes it a really useful resource.

I do this with medical test results too; it clarifies and explains all the technical medical terms.
 
An LLM has nothing at stake.

I rely on one a lot. Every month I pay out of pocket in the three-digit range for a pro subscription from one of the big providers. That same tool has proposed actions that would have been completely ruinous to the systems I handle.

I am not saying the platform should be disregarded. My point is that relying on it while you are unsure about when to disagree can lead into risky territory. The LLM has no concern either way.
 
For my lab reports from Trust Pointe I go with Grok. Those come with far more detail and a lot more testing. Does a great job.
 
AI can be a valuable tool, yet I wouldn't rely on it for anything I don't grasp. Its design uses a reward mechanism, one that incentivizes pleasing the user—even when that means giving wrong answers. I wouldn't put complete faith in it. Ask lots of questions and verify the sources it drew from. It trains on the Internet, and everyone knows how much fake nonsense is out there. All it does is spit that same information back out.
 
One of the easiest ways to cut down on that problem is to use this prompt: " do not agree with me without providing evidence " It isn't perfect. Still, when you pair it with research/scholar mode in chatgpt, the responses tend to be far more scientifically sound, and the model's natural pull to side with you and spin out believable-sounding tales rather than stick to facts is weakened. On GLP1Chat, I've come across a number of outputs from llms that were quite poor, leaning much more toward bro science than actual research. This method, at least, is simple and straightforward, and it does provide references that are usually not made up and are easy to verify.
 
I share ChatGPT's assessment here:

"Based on the Facebook profile link alone, I would be very cautious."

Given that you've been on GLP-Forum for some time, I'd suggest beginning your vendor search in this community, looking closely through the threads and feedback here.

Steer clear of vendors on Facebook.
 
Back
Top