ChatGTP

[redbullzuiper]

I was checking what the knownledge of ChatGTP was regarding pyrotechnics. Well, hes still an amateur ^^

 

 

A common pyrotechnic composition that utilizes potassium nitrate is a whistle mixture. Here is a possible whistle mixture composition using potassium nitrate:

Potassium nitrate (oxidizer): 70%

Charcoal (fuel): 20%

Sulfur (fuel): 10%

This composition can be used to create a whistling effect when it is burned. The potassium nitrate acts as the oxidizer, which provides the oxygen needed for the charcoal and sulfur to burn. The charcoal and sulfur act as fuels, which burn and create the whistling sound.

When this mixture is packed into a tube, and ignited, it will burn rapidly and produce a loud whistling sound as the gases escape through the tube. Keep in mind that this is a simplified recipe and this mixture should be handled with care, as potassium nitrate is a powerful oxidizer and can be explosive if not handled properly. It should be stored in a cool, dry place, and should not be allowed to come into contact with fuels or other combustible materials.

 

What about a coloured composition based on AP?

 

 

Here is another possible green star composition using ammonium perchlorate:

Ammonium perchlorate (oxidizer): 80%

Barium chlorate (colorant and oxidizer): 10%

Aluminum powder (fuel): 5%

Shellac (binder): 5%

As in the previous composition, the barium chlorate is the chemical that gives the green color when it burns, it also acts as an oxidizer. The ammonium perchlorate and barium chlorate together will provide enough oxygen to burn the aluminum powder, which acts as the fuel. The shellac is the binder that holds the mixture together.

Again, it's important to note that this is just one example of a green star composition, and the specific proportions of each ingredient may be adjusted depending on the desired effect, or other variations of the compounds can be used.

 

What about flash powder?

 

 

Here is a possible flash powder composition using potassium nitrate:

Potassium nitrate (oxidizer): 75%

Aluminum powder (fuel): 25%

This composition is a basic flash powder, where Potassium nitrate acts as the oxidizer and aluminum powder acts as the fuel. It is important to note that this is a very basic flash powder composition, and as such it is not recommended for any kind of use, as it may be extremely dangerous.

 

Flash powder based on KClO4?

 

 

A possible flash powder composition is:

Potassium perchlorate (oxidizer): 70%

Aluminum powder (fuel): 20%

Iron powder (fuel): 10%

 

I wonder where it gets its data from.

[JAWChemist]

In APC Pyro's Discord chat, ozarkpyro tried out chatGPT.

It seems it's in his field of expertise (DM), doubts it is trustworthy and mused on whether it could be train in the field specifically.

He got interesting suggestions on shell construction. He also quizzed it on blue compositions and shared the results.

I had coincidental tried the same thing (making blue color in pyrotechnics) but the results were so poor they were not even worth noting.

[cmjlab]

I had to look up ChatGTP to see what it was, funny enough, in the descriptions it says that one of it's main flaws is factually incorrect responses. But it apparently is excellent at writing forum response amongst other things! It reminds me of a couple people who used to post here before they were banned (they could similarly mimic educated forum responses, but were factually inaccurate on almost all of the information provided) :-) Edited January 19 by cmjlab

[JAWChemist]

Their server is overloaded and it is difficult to get a history of your questions. I would like to review my blue color question.

 

Dilbert https://dilbert.com/ has a series of panels (~4) that highlight issues with AI. The 1/17/23 panel is interesting. My son (educated in the area) said it took the internet about 0.1s to get around chatGPT ethical filter and travel time problems (1 car 1 minute to go 1 mile, how long will it take 4 cars - gave the wrong answer). He developed some interesting scifi/fantasy plot lines. I got a sci-fi plot line developed on bismuth but now I need to learn how to write a story effectively.

 

Well to tie it to the overall forum; bismuth to bismuthates to blue color/oxidizer like Cu arsenite to xenates (outrageously extreme to provoke a conversation about ‘nates/’nites in another thread but it got dissed. Am I KEWL if I use dissed for disrespected (rhetorical)?).

[cmjlab]

I'm confused, are you asking me JAW? I don't see a question about blue color.

[JAWChemist]

Sorry for the confusion.

 

The "question" comment was a statement about a previously asked question for chatGPT on blue color. I would have liked to review the chatGPT answer to my blue question for the discussion but could not because chatGPT was overloaded. It would have been to show the quality of its responses as part of the discussion.

 

The one statement that has a "?" is a humorous rhetorical question and no response is expected.

 

I just finished a comment on the "Interesting BP blog" topic (Ulrich Bretscher) that included a reference to Roger Bacon.

More humor :-) Talk to you on the flip side of the Bacon.

[Sulphurstan]

Hi Redbullzuipper.

I sent you a PM ... ;-)

[Arthur]

The trouble is that it's reference material is "the internet" some of which is 100% BS and no AI program can detect that.

 

Someone has been trying to market a speech recognition version to give real time advice in court so that defendants can be without a lawyer, the trouble was that the system understood the major parts of law but got the specific details totally wrong.

[ozarkpyro]

This is true, and a general purpose deep knowledge platform will likely struggle with things like this. I do wonder if a system could be trained with some hard experimental data - stuff like composition, burn time, temperature, spectral emissions, smoke production, and anything else one could possibly record...

 

Perhaps it could, at first, be used to fine tune some formulas out there as balanced stoichiometry isn't everything. Eventually I could foresee it developing insights with regards to chemical interactions and spectral emissions that would be hard to get at otherwise.

 

Later it could be trained with various other chemistry research papers to find other potential compositions - I'm looking at you Bromine!

[Arthur]

With the fact that most pyro comps were invented and developed in the days before internet, and that many people copy them wrongly and transcribe their errors onto the internet, I'd have little faith in any purely internet research tool finding safe and correct answers.

[JAWChemist]

It's not rocket science. https://www.npr.org/2023/02/02/1152481564/we-asked-the-new-ai-to-do-some-simple-rocket-science-it-crashed-and-burned

AI researcher Gary Marcus, "are just autocomplete on steroids."

[projectawry]

This really makes you wonder how many unintended disasters are going to be caused by incorrect AI information.

[SharkWhisperer]

This really makes you wonder how many unintended disasters are going to be caused by incorrect AI information.

The opening thread has a green comp that contains ammonium perchlorate and barium chlorate. Does this seem like a wise idea?

 

Please do not rely on ChatGPT for pyro information. It is just not reliable. at least not yet. And it guesses, often horribly. Last night it could not even put together an accurate solubility table for NaCl in water that accounts for volume contraction, ffs, even after 8 or 9 attempts with clarified instructions. Instead of saying it did not have adequate information it just made wrong stuff up.

[Arthur]

A US Lawyer has just been reprimanded by a judge for filing a brief created by CHatGPT that contained so many errors that it wasn't valid and had FALSE citations quoted. Chatbots cannot be relied on.