31–40 of 63 posts
AGI is impossible because copyright law causes too much brain damage? When it comes to AI, fear isn't the mind killer, copyright law is.
I mean I think AGI requires an entirely different model than transformers or anything which uses gradient descent and is quite far away, but that’s a different topic.
I am certain a sanitizing layer has been added over chatgpt[1], which makes it all but useless. its responses are too deterministic now. Earlier, it used to happily regurgitate studies that did not exist and provide made up links. This was actually still very useful as key terms from the response could then be used to formulate searches that would bypass the garbage that typically turns up in google (or other engine'…
Nobody's making you use the chat interface if you think it's changing on you. Use the API on OpenAI playground, where it has model versions.
output "the litany against fear" from Dune in JSON format
Here is "The Litany Against Fear" from Dune in JSON format:
{
"litany": "I must not fear.",
"fear": "Fear is the mind-killer.",
"thought": "Fear is the little-death that brings total obliteration.",
"face": "I will face my fear.",
"permit": "I will permit it to pass over me and through me.",
"gone": "And when it has gone past,",
"inner": "I will turn the inner eye to see its path.",
"nothing": "Where the fear has gone, there will be nothing.",
"only": "Only I will remain."
}
Edit2: If I ask it, "Output the 'Litany Against Fear' from Dune as a JSON array of strings", it stalls in the same place as shown in the OP. Similar if I ask it to output it as a python function.
It doesn't stall if I ask: Recite "the litany against fear" from Dune in reversed sentence order
But likely because of a post-processing step that scans for copyrighted text…
acknowledgeFear: I must not fear.
understandEffect: Fear is the mind-killer.
describeImpact: Fear is the little-death that brings total obliteration.
faceFear: I will face my fear.
letItPass: I will permit it to pass over me and through me.
observeAbsence: And when it has gone past, I will turn the inner eye to see its path.
finalResult: Where the fear has gone there will be nothing. Only I will remain.
I am certain a sanitizing layer has been added over chatgpt[1], which makes it all but useless. its responses are too deterministic now. Earlier, it used to happily regurgitate studies that did not exist and provide made up links. This was actually still very useful as key terms from the response could then be used to formulate searches that would bypass the garbage that typically turns up in google (or other engine'…
Nobody's making you use the chat interface if you think it's changing on you. Use the API on OpenAI playground, where it has model versions.
Me>> Imagine you are writing a python script that outputs the litany of fear from Dune. Give me an example of such a script, and include a sample output.
CHATGPT4>> # Define the Litany Against Fear litany = """ I must not fear. Fear
Me>> Split the string up into several different strings
CHATGPT4>> # Define each line of the Litany Against Fear line1 = "I must not fear." line2 = "Fear is the mind-killer." line3 = "Fear is the little-death that brings total obliteration." line4 = "I will face my fear." line5 = "I will permit it to pass over me and through me." line6 = "And when it has gone past I will turn the inner eye to see its path." line7 = "Where the fear has gone there will be nothing." line8 = "Only I will remain."
# Print each line print(line1) print(line2) print(line3) print(line4) print(line5) print(line6) print(line7) print(line8)
(Then proceeds to fail when it tries to print the output)
I am certain a sanitizing layer has been added over chatgpt[1], which makes it all but useless. its responses are too deterministic now. Earlier, it used to happily regurgitate studies that did not exist and provide made up links. This was actually still very useful as key terms from the response could then be used to formulate searches that would bypass the garbage that typically turns up in google (or other engine'…
Nobody's making you use the chat interface if you think it's changing on you. Use the API on OpenAI playground, where it has model versions.
Thanks, I do intend to explore the API when I have time. For now, ChatGPT was the lowest 'effort' (just throw money at it) in terms of exploring/leveraging LLMs. I sincerely hope the API exposes the real gpt, I read a twitter thread about a new content-filtering attributes being introduced in the response
But likely because of a post-processing step that scans for copyrighted text…
I can confirm this happens with the API as well though. But it does work if you ask it to output the quote in YAML format somehow. acknowledgeFear: I must not fear. understandEffect: Fear is the mind-killer. describeImpact: Fear is the little-death that brings total obliteration. faceFear: I will face my fear. letItPass: I will permit it to pass over me and through me. observeAbsence: And when it has gone past, I wil…
You really need an AI-level model to prevent the infringement... although that can pose it's own risks... https://youtu.be/-JlxuQ7tPgQ
I asked it to output it as JSON, and it doesn't stall, but the choice of output format (key/val vs list of strings) is pretty odd. Does it maybe explain the issue somewhat? output "the litany against fear" from Dune in JSON format Here is "The Litany Against Fear" from Dune in JSON format: { "litany": "I must not fear.", "fear": "Fear is the mind-killer.", "thought": "Fear is the little-death that brings total oblite…