The Masonic Egyptian Temple Endowments

[Calm]

37 minutes ago, Analytics said:

r. That conversation is what inspired me to use Chat GPT to respond to PacMan

It is about what gets put into the AI as well as what is done with it.  I can imagine the accounting is worth that much because of the time it took to train that particular AI and its programmed checks to avoid errors probably was a big part of that.  ChatGPT is problematic because it has more open sources as I understand, less vetting on what goes in.

Edited August 9, 2024 by Calm

[webbles]

46 minutes ago, Calm said:

It is about what gets put into the AI as well as what is done with it.  I can imagine the accounting is worth that much because of the time it took to train that particular AI and its programmed checks to avoid errors probably was a big part of that.  ChatGPT is problematic because it has more open sources as I understand, less vetting on what goes in.

Most companies don't train a new AI.  That requires a lot of money, time, and effort.  Instead you customize the "prompts".  Every time you interact with ChatGPT or related, there is a prompt that is automatically fed into it.  Depending on the prompt, the AI acts differently.  You can take the exact same AI and give it two different prompts and they can respond 100% differently.  It has been a bit of a challenge finding the correct prompt for the specific need and that is where there are "experts" who you can pay a lot of money to design your prompts.  And as ChatGPT is updated, sometimes the prompts are no longer as useful.

[PacMan]

2 hours ago, ZealouslyStriving said:

Because the Endowment of the House of the Lord is patterned after (higher) pre-Israelite religion, not the (lesser) Law of Moses.

Bingo. 

[Okrahomer]

4 hours ago, Analytics said:

I will state for the record that my prompt was geared to making Chat GPT answer this question as objectively as possible, with the context of whether the similarities and alleged similarities between Egyptian temples and Mormon temples are so similar that an explanation for that similarity is needed.  

In my line of work, I am frequently called upon to provide guidance with respect to legal and regulatory requirements for the food/dietary supplement industry.  On more than one occasion, I have been asked a question for which I had only a vague or incomplete notion of the correct answer.  Digging down to a correct and supportable answer takes a lot of research.

A recent example was a question similar to this one:  “Is cGMP required for an investigational dietary supplement to be used in a Clinical Trial?”

My gut told me the answer would be “no”, but I was not at all sure; so I started reading relevant CFRs and Guidance documents.  After hours of research— finding no definitive supporting regulation(s)— my tired and lazy brain said:  Hey, Kevin!  Why don’t you ask ChatGPT?  So I plugged the exact question in, and I got something like this in response:

“Yes, cGMP is required for an investigational dietary supplement to be used in a Clinical Trial.”  

I asked ChatGPT to provide chapter and verse.  It gave me chapter and verse from all the same CFRs I had just read — none of which was at all definitive and in some cases were simply irrelevant.

So it occurred to me to use a series of fundamental questions, basically leading ChatGPT through the same analysis I had already done on my own.  Eventually ChatGPT responded to the original question like this: 

“No.  There is no regulation that requires cGMP for an Investigational food supplement to be used in a Clinical Trial.”

I have learned that ChatGPT often responds in purely conjectural and inaccurate ways, but does not say so! My take away:  Anyone who relies on ChatGPT for accurate/correct/unbiased information does so at their own peril.

Edited August 10, 2024 by Okrahomer

[Calm]

40 minutes ago, Okrahomer said:

but does not say so

This is the problem I have with it.  Otherwise I think it’s great.

[Okrahomer]

6 minutes ago, Calm said:

This is the problem I have with it.  Otherwise I think it’s great.

I was just going to edit to say that!  It can be quite useful, but it really needs to indicate when it’s not sure or is providing only a best guess.