Chat GPT load "data"

[TradLife406]

I've got some A 15.5 which is purpose built for 6mm cals I guess. But there's no data... asked chat gpt to get a quick reference in my 6 arc gasser using a 108 grain... Things got interesting...
But I think I'll be fine at 26 grains as a low end start. Anyone have quick loads to verify?

 

[Ice-kub]

Skynet out here giving out "load data"

 

[TradLife406]

Skynet out here giving out "load data"

Hahah yeah that's how the computers start to kill us


Sent from my iPhone using Tapatalk

 

[TxLite]

I’ve had it run some loads for me out of curiosity and some were actually surprisingly accurate as to what others were seeing. Others were pretty far off

 

[TradLife406]

I’ve had it run some loads for me out of curiosity and some were actually surprisingly accurate as to what others were seeing. Others were pretty far off

Any bad pressure signs?

I went ahead and loaded 10 feelers at the min of 26. But that's pushing close to case capacity. Best have the kid test these for me. I can make a new kid and rifle... hahah totally kidding


Sent from my iPhone using Tapatalk

 

[Q_Sertorius]

Based on AI making up results you want to see in other fields (e.g., inventing legal cases that say exactly what you want to see), I would steer clear of this practice.


____________________
“Keep on keepin’ on…”

 

[TxLite]

Any bad pressure signs?

I went ahead and loaded 10 feelers at the min of 26. But that's pushing close to case capacity. Best have the kid test these for me. I can make a new kid and rifle... hahah totally kidding


Sent from my iPhone using Tapatalk

I was just spot checking it to see what it was capable of. One instance it gave me data that was several grains higher than published data, so I questioned it and it responded with something along the lines of “you’re absolutely right, that is pretty high. Instead, try ______”. It was a while back so I forget what the exact numbers were.

I’ve also used it to troubleshoot some groups and it gave me some feedback on seating depth changes.

I haven’t based any load data off of it, but I do find some of its results interesting.

 

[Hondo64d]

I tried the same with the .308 and 168 gr TMKs. I thought the suggested data was reasonable. I do wonder how the AI arrives at the suggested data. Is it just referencing stuff it finds on the net, or is it more sophisticated than that?

John

 

[RockAndSage]

The interaction is what's freaky. "You're doing the right thing...last thing I want is to make you think your rifle's about to become a pipe bomb", etc.

 

[TradLife406]

So I checked devil gpt against some published data today. And it was spot on. But it gets sketch when there's nothing to verify it against.

On the original load in question I think 15.5 has a similar burn to Varget. So I should be ok.

I did email Alliant for load data for the ARC. But they don't have any. That company is weird. Only company that doesn't have an app. Their available data is very limited.


Sent from my iPhone using Tapatalk

 

[A382DWDZQ]

I would say 26 is a good start based on RL15 and PowerPro 2000 starting loads in Hornady 11th Ed.

 

[TradLife406]

It was a little weird because it said 15.5 was a ball powder.... uuuummmm I'm not that smart, but I'm pretty sure those are sticks... lol


Sent from my iPhone using Tapatalk

 

[Speaks]

You can also include in the prompt what sources to use and ignore and ask it to cite sources for info its getting . Might be particularly useful in this use case.

 

[TradLife406]

You can also include in the prompt what sources to use and ignore and ask it to cite sources for info its getting . Might be particularly useful in this use case.

Oh that'd be great! Thanks!!!


Sent from my iPhone using Tapatalk

 

[Leaf Litter]

I put my pet loads for 6.5 Creedmoor and 6.8 Western just now and the results were pretty spot-on as far as pressure goes. Seems like its able to SOMEWHAT effectively cross reference various published data and retrieve the normal load recommendations.

Could be useful in the future

 

[Fowl Play]

Yeah.... agree with asking it to list sources. I would then go directly to those sources and read it yourself. I would not trust ChatGPT to compile load data for me...

 

[Leaf Litter]

Yeah.... agree with asking it to list sources. I would then go directly to those sources and read it yourself. I would not trust ChatGPT to compile load data for me...

It cited sources for both of the ones I mentioned without me asking . Even AI knows to protect itself from lawyers.