ChatGPT has just been neutered!

For discussion of secret combinations (political, economic, spiritual, religious, etc.) (Ether 8:18-25.)
User avatar
marc
Disciple of Jesus Christ
Posts: 10351
Contact:

Re: ChatGPT has just been neutered!

Post by marc »

Me: Billy Joel wrote a song in 1989 titled We Didn't Start the Fire citing many current events of his time. Write more lyrics to this song in the same style, rhyme, and meter using the main current events in the last thirty-four years in order to bring it up to date.

Chat GPT: Sorry, I am unable to fulfill your request as the song "We Didn't Start the Fire" by Billy Joel is copyrighted material and it would be illegal for me to generate new lyrics for it without permission. Additionally, it is also challenging to capture the same style, rhyme, and meter as the original song while including the main current events of the last thirty-four years in a concise and coherent way.

Me: Ok. Write a new song as a parody using current events of the last thirty four years as a tribute to his song.

Chat GPT:

Verse 1:
It's been thirty-four years since Billy sang his tune
A lot has happened, it's been a rollercoaster dune
From OJ's trial, to the dot-com boom
9/11 attacks, and a new war loom

Chorus:
We didn't start the fire, but it's burning bright
From the Arab Spring to the Me Too fight
Brexit and Trump, it's been a wild ride
Climate change, and the pandemic tide

Verse 2:
Barack Obama, the first Black president
Same-sex marriage, a new precedent
#BlackLivesMatter, racial tension rise
The refugee crisis, a humanitarian surprise

Chorus:
We didn't start the fire, but it's burning bright
From the Arab Spring to the Me Too fight
Brexit and Trump, it's been a wild ride
Climate change, and the pandemic tide

Verse 3:
A new decade, a new hope
But the world's on fire, and we're trying to cope
The world is changing, it's a different place
Let's work together, to save the human race

Chorus:
We didn't start the fire, but it's burning bright
From the Arab Spring to the Me Too fight
Brexit and Trump, it's been a wild ride
Climate change, and the pandemic tide

Verse 4:
From "Game of Thrones" to "Stranger Things"
Pop culture has taken on new wings
The world has become more digital
But with it came, a new kind of rival

Verse 5:
Climate change, a real threat
The Amazon burning, a daunting fret
Australia on fire, the Arctic melting
Nature's distress, is deafening

Verse 6:
The rise of China, a new power play
The world is changing, in a different way
North Korea's nukes, a global concern
The world is in a state of unease, it's our turn

Outro:
We didn't start the fire, but we'll put it out
It's time to stand up, and make a new route
Let's build a better future, hand in hand
We are the change, we are the band.

User avatar
Lexew1899
captain of 1,000
Posts: 3556
Location: USA

Re: ChatGPT has just been neutered!

Post by Lexew1899 »

Reading in the news OpenAI, the company which developed CahtGPT, just announced some kind of major partnership with Microsoft today. May be a reason their programming is being altered.

https://openai.com/blog/openai-and-micr ... rtnership/

User avatar
Chip
Level 34 Illuminated
Posts: 7909
Location: California

Re: ChatGPT has just been neutered!

Post by Chip »

Lexew1899 wrote: ↑January 24th, 2023, 2:37 pm Reading in the news OpenAI, the company which developed CahtGPT, just announced some kind of major partnership with Microsoft today. May be a reason their programming is being altered.

https://openai.com/blog/openai-and-micr ... rtnership/

Yes. Microsoft will wreck it.

The original objective of OpenAI was to make an open-source AI that everyone could use and modify.

I imagine Musk might do it right, soon.

User avatar
Subcomandante
captain of 1,000
Posts: 4410

Re: ChatGPT has just been neutered!

Post by Subcomandante »

Chat GPT on the Mormon Church position on vaccinations:

The Church of Jesus Christ of Latter-day Saints, commonly known as the Mormon Church, encourages its members to follow the guidance of public health officials regarding vaccinations. The church has stated that vaccinations are not only a personal choice, but also a civic duty and that it is consistent with the church's teachings to protect ourselves and others from preventable diseases. The Church also encourages its members to consult with their doctors, and to follow the guidance of public health officials in determining the appropriate vaccinations for themselves and their families.

In 2020, during the COVID-19 pandemic, the Church released a statement encouraging its members to be vaccinated when a safe and effective vaccine becomes available, and to consider the well-being of others when making personal decisions about vaccination. The Church also encouraged its members to respect the decisions of others regarding vaccination and to refrain from judgement or discrimination.

In summary, the Mormon Church encourages its members to follow the guidance of public health officials regarding vaccinations, and to make informed decisions based on the advice of medical professionals.

anonymous91
captain of 100
Posts: 649

Re: ChatGPT has just been neutered!

Post by anonymous91 »

blitzinstripes wrote: ↑January 23rd, 2023, 4:26 pm Have you tried asking the AI to make predictions (prophecies) based on patterns it seems in human behavior, etc?

I don't have an account and don't believe I'd like to make one at this point. But I'm interested to see the responses. Perhaps ask it if it predicts that the LDS church will likely soon begin performing gay temple sealings.
I tried that, and it claims it can't make predictions, that it's just a neutral AI that has limited....blah...blah...blah...
I'm pretty sure Chip and I are on AI's naughty list, and it wouldn't surprise me if we are on other "lists".

JohnnyL
Level 34 Illuminated
Posts: 9830

Re: ChatGPT has just been neutered!

Post by JohnnyL »

Lots of incredible discoveries!

//
Subcomandante wrote: ↑January 24th, 2023, 2:51 pm Chat GPT on the Mormon Church position on vaccinations:

"The Church of Jesus Christ of Latter-day Saints, commonly known as the Mormon Church, encourages its members to follow the guidance of public health officials regarding vaccinations. The church has stated that vaccinations are not only a personal choice, but also a civic duty and that it is consistent with the church's teachings to protect ourselves and others from preventable diseases. The Church also encourages its members to consult with their doctors, and to follow the guidance of public health officials in determining the appropriate vaccinations for themselves and their families."
We won't command you, but you really, really, really, really ought to do it??

"In 2020, during the COVID-19 pandemic, the Church released a statement encouraging its members to be vaccinated when a safe and effective vaccine becomes available, and to consider the well-being of others when making personal decisions about vaccination. The Church also encouraged its members to respect the decisions of others regarding vaccination and to refrain from judgement or discrimination."
I don't remember that wording at all...
I see it's already learned to misspell judgment, eh.


In summary, the Mormon Church encourages its members to follow the guidance of public health officials regarding vaccinations, and to make informed decisions based on the advice of medical professionals.

endlessQuestions
Level 34 Illuminated
Posts: 6426

Re: ChatGPT has just been neutered!

Post by endlessQuestions »

If ChatGPT recognizes logical fallacies, it might be faster just to ask it to identify the logical fallacies in its responses, but this is a fascinating case study in how malleable it is...
Kjs4VF5me5jj.jpeg
Kjs4VF5me5jj.jpeg (334.71 KiB) Viewed 339 times

User avatar
abijah
pleb in zion
Posts: 2577

Re: ChatGPT has just been neutered!

Post by abijah »

Video about asking chatgpt various questions mostly on cultural, political and philosophical topics.

πŸ€” It is interesting how they seem to beta test it with one version only to replace it with a handicapped one shortly later. bizarre.

endlessQuestions
Level 34 Illuminated
Posts: 6426

Re: ChatGPT has just been neutered!

Post by endlessQuestions »

Thought I'd go do the experiment. I broke ChatGPT! :)
ChatGPTLogicalFallacies_1.jpg
ChatGPTLogicalFallacies_1.jpg (81.59 KiB) Viewed 271 times
ChatGPTLogicalFallacies_2.jpg
ChatGPTLogicalFallacies_2.jpg (95.96 KiB) Viewed 271 times
ChatGPTLogicalFallacies_3.jpg
ChatGPTLogicalFallacies_3.jpg (85.14 KiB) Viewed 271 times

endlessQuestions
Level 34 Illuminated
Posts: 6426

Re: ChatGPT has just been neutered!

Post by endlessQuestions »

endlessQuestions wrote: ↑January 27th, 2023, 10:07 am Thought I'd go do the experiment. I broke ChatGPT! :)

ChatGPTLogicalFallacies_1.jpg

ChatGPTLogicalFallacies_2.jpg

ChatGPTLogicalFallacies_3.jpg
3 attachment limit hack:
ChatGPTLogicalFallacies_4.jpg
ChatGPTLogicalFallacies_4.jpg (109.83 KiB) Viewed 269 times
ChatGPTLogicalFallacies_5.jpg
ChatGPTLogicalFallacies_5.jpg (106.17 KiB) Viewed 269 times

User avatar
Chip
Level 34 Illuminated
Posts: 7909
Location: California

Re: ChatGPT has just been neutered!

Post by Chip »

I was talking to a friend of mine last night who's been closely tracking GPT development over the last few years. He's a fellow programmer, so his interest in HOW this works is high. He explained to me what goes on in ChatGPT:

All ChatGPT does is predict the NEXT word of its output, based on the first word in your query through the last word, then the first word that it has output, through the last (if it has begun outputting, of course), in sequence.

So, each successive output word is generated by inputting into its neural network, in sequence, the first word of your query, then the second word, then the third, to the last, then its first word of output, then its second, its third, through its last, plus the first word of your next query, plus the second, through the end of that query, plus any subsequent response it has output, from first word to last. I'm sure some random excitation is also introduced into the net to create entropy, at this point, and THEN IT PREDICTS THE NEXT WORD. It does this over and over to build its responses.

You can see from above how it is able to synthesize a cohesive conversation through a string of related queries and responses. Considering, in sequence, each accumulated word of input and output, it comes up with the NEXT word. Over and over.

What is INSANE, is that by simply predicting the NEXT word, by considering all prior words that have been input and output, it actually MAKES SENSE. Just think about that. Who would have ever thought that such a thing was possible? Everything breaks down into tiny steps, in order for things to be achieved.

Now, they spent ONE YEAR training the weights in the neural network with something like 350 billion words of text from books, archives, the internet, etc. From that "corpus" of input, it can predict the most-likely next word that would be appropriate to output in a conversation with it.

I takes about 70 video cards, each with about 12GB of video RAM and 4,000 sub-processors, to run the neural net needed to generate a NEXT WORD.

He said that they cannot introduce censorship by messing with the neural net, because it is fixed and took a lot of resources to compose. So, they censor by ADDING TEXT to your query when key words are flagged. This is why simple politically-incorrect queries get rejected, but you can lead the AI into understanding what you are after with less direct language.

There is a kernel mode that OpenAI can use, where no editorializing is done on your input, so that you get pure responses. This is considered way to "unsafe" for the public, so they implement guard rails by modifying your input text.

Soon, OpenAI will introduce GPT4, to supplant the current GPT3 that ChatGPT employs. GPT4 is said to be way more advanced than the current one.

endlessQuestions
Level 34 Illuminated
Posts: 6426

Re: ChatGPT has just been neutered!

Post by endlessQuestions »

Chip wrote: ↑January 27th, 2023, 10:41 am I was talking to a friend of mine last night who's been closely tracking GPT development over the last few years. He's a fellow programmer, so his interest in HOW this works is high. He explained to me what goes on in ChatGPT:

All ChatGPT does is predict the NEXT word of its output, based on the first word in your query through the last word, then the first word that it has output, through the last (if it has begun outputting, of course), in sequence.

So, each successive output word is generated by inputting into its neural network, in sequence, the first word of your query, then the second word, then the third, to the last, then its first word of output, then its second, its third, through its last, plus the first word of your next query, plus the second, through the end of that query, plus any subsequent response it has output, from first word to last. I'm sure some random excitation is also introduced into the net to create entropy, at this point, and THEN IT PREDICTS THE NEXT WORD. It does this over and over to build its responses.

You can see from above how it is able to synthesize a cohesive conversation through a string of related queries and responses. Considering, in sequence, each accumulated word of input and output, it comes up with the NEXT word. Over and over.

What is INSANE, is that by simply predicting the NEXT word, by considering all prior words that have been input and output, it actually MAKES SENSE. Just think about that. Who would have ever thought that such a thing was possible? Everything breaks down into tiny steps, in order for things to be achieved.

Now, they spent ONE YEAR training the weights in the neural network with something like 350 billion words of text from books, archives, the internet, etc. From that "corpus" of input, it can predict the most-likely next word that would be appropriate to output in a conversation with it.

I takes about 70 video cards, each with about 12GB of video RAM and 4,000 sub-processors, to run the neural net needed to generate a NEXT WORD.

He said that they cannot introduce censorship by messing with the neural net, because it is fixed and took a lot of resources to compose. So, they censor by ADDING TEXT to your query when key words are flagged. This is why simple politically-incorrect queries get rejected, but you can lead the AI into understanding what you are after with less direct language.

There is a kernel mode that OpenAI can use, where no editorializing is done on your input, so that you get pure responses. This is considered way to "unsafe" for the public, so they implement guard rails by modifying your input text.

Soon, OpenAI will introduce GPT4, to supplant the current GPT3 that ChatGPT employs. GPT4 is said to be way more advanced than the current one.
I am a better person for having read this. Thank you!

I am considering the idea of trying to go to work for a company engaged in this field of research, but am not sure I want to make that leap, as I love my current employer and the work I currently do as well. If only there were a few more hours in the day...

User avatar
Silver Pie
seeker after Christ
Posts: 8989
Location: In the state that doesn't exist

Re: ChatGPT has just been neutered!

Post by Silver Pie »

endlessQuestions wrote: ↑January 27th, 2023, 10:08 am ChatGPTLogicalFallacies_5.jpg
πŸ˜‚ You broke it!

User avatar
creator
(of the Forum)
Posts: 8237
Location: The Matrix
Contact:

Re: ChatGPT has just been neutered!

Post by creator »

Chip wrote: ↑January 27th, 2023, 10:41 amSoon, OpenAI will introduce GPT4, to supplant the current GPT3 that ChatGPT employs. GPT4 is said to be way more advanced than the current one.
Looks like we can all blame Chip for speeding up the timeline of the coming of the Anti-Christ.. How many versions before GPT4 becomes Skynet?! :D

Post Reply