Skip to main content

Search

Items tagged with: ChatGPT


An open-source complete PyTorch-based #ChatGPT equivalent implementation process: https://www.hpc-ai.tech/blog/colossal-ai-chatgpt

I need more hours in the day to play with this stuff.


Dziś w #OkoPress mój tekst o #ChatGPT i "sztucznej inteligencji":

https://oko.press/chatgpt-cala-prawda-o-wielkich-modelach-jezykowych

> Antropomorfizacja modeli uczenia maszynowego, takich jak ChatGPT, ma na celu przekonanie nas, że nawet jeśli te technologie nie są całkowicie bezpieczne i nieszkodliwe, to są przynajmniej neutralne. Po to, by trudniej było nam dostrzec, jaką mogą wyrządzać krzywdę

> Warto przyjrzeć się bliżej zakodowanym w nich uprzedzeniom i temu, komu służą — a kogo mogą krzywdzić.

1/🧵


For the first time we got an inquiry by email asking us about completely made-up claims about our organization which were generated via .. #ChatGPT.

I'm sure we're not the only ones. So yeah, the misinformation these models hallucinate already has real-world consequences, probably including more severe ones than that.

The warnings ChatGPT displays to new users are inadequate -- and the integration of similar models in search engines is reckless and premature.


Reminder that "free research access" to OpenAI's #ChatGPT really means *you* are giving *them* free access to human intelligence for *their* research

(You don't own your prompts; have no access to your prompt history; cannot erase them; and have consented to release all your prompt tuning cleverness to them in perpetuity — all in exchange for the privilege of developing a serious dependency on text regurgitation that will form the backbone of their soon to be introduced subscription model)



Given what we've learned about and from #ChatGPT, do you think it more or less likely that we'll see a useful "Young Ladies Illustrated Primer" in our lifetimes? (See: https://en.wikipedia.org/wiki/The_Diamond_Age)
@evan

  • More likely (60%, 3 votes)
  • Less likely (40%, 2 votes)
  • Not sure (0%, 0 votes)
  • I've never read "The Diamond Age" (0%, 0 votes)
5 voters. Poll end: 1 year ago


The hype around AI tools like ChatGPT and Stable Diffusion is in overdrive right now, but what is its actual impact likely to be?

This week on #TechWontSaveUs, I chat with @timnitGebru about the misleading framing of AI, why she started doing this research, and how she ended up getting fired by Google. It’s a must listen!

Tune in here: https://www.buzzsprout.com/1004689/12073262-don-t-fall-for-the-ai-hype-w-timnit-gebru

#ai #tech #podcast #chatgpt


This is quite the exposé!

I knew that OpenAI has been using crowd workers for at least the last two years to build instruct-tuning datasets for #ChatGPT. I didn’t know they were also outsourcing violence, hate speech, and sexual abuse labeling to Kenyans for $2/hour.

The work is apparently so traumatic that the outsourcing company, Sama, has cut ties with OpenAI.

https://time.com/6247678/openai-chatgpt-kenya-workers/


Well, fuck. The GPT disinformation age is now.

I googled "OS for 4gb ram" and the first hit, which also was used by google to populate its snippet is an answer from quora which is very obviously created with #chatgpt (I recognised the non-committal non-answer right away, but it can also be detected by a popular GPT detector).

The user has 98 answers and, you've guessed it, they are all created with GPT.
PLOT TWIST: The questions where also created with GPT!

https://www.quora.com/profile/Heri-Mulyo-Cahyo

#AI
google spinet showing an answer from quora: 
"Windows 10: Windows 10 is the latest version of the Windows operating system, and it is designed to be efficient and versatile. It includes features like Windows Hello and Cortana, which can help users get more done faster, as well as tools like Task Manager, which can help manage memory usage."
A quora answer to the question "What OS should be good for a 4 GB RAM PC/laptop?"


"For a PC or laptop with 4 GB of RAM, it is recommended to use an operating system that is optimized for performance and can efficiently manage the available memory. Some popular operating systems that would be suitable for a 4 GB RAM PC or laptop include:

    Windows 10: Windows 10 is the latest version of the Windows operating system, and it is designed to be efficient and versatile. It includes features like Windows Hello and Cortana, which can help users get more done faster, as well as tools like Task Manager, which can help manage memory usage.
    macOS: macOS is the operating system used by Apple's Mac computers. It is known for its user-friendly interface, robust security, and access to a wide range of apps and services. macOS also includes tools like Activity Monitor, which can help users monitor and optimize their memory usage.
    Linux: Linux is an open-source operating system that is widely used on servers and other high-performance systems. It is known for its flexibility, stability, and security, and it can be customized to meet a wide range of needs. Many Linux distributions, such as Ubuntu and Fedora, are optimized for use on low-memory systems.

Ultimately, the best operating system for a 4 GB RAM PC or laptop will depend on the specific needs and preferences of the user. It is important to research and compare the different options before making a decision."


ChatGPT gets treated like technological magic, but that ignores the humans behind the curtain that make it function.

OpenAI paid Sama to hire Kenyan workers at $1.32 to $2 an hour to review “child sexual abuse, bestiality, murder, suicide, torture, self harm, and incest” content. Their work made the tool less toxic, but left them mentally scarred. The company ended the contract when they found out TIME was digging into their practices.

https://time.com/6247678/openai-chatgpt-kenya-workers/

#tech #chatgpt #ai


In this interview, the author explains she got ChatGPT to spit out writing in the style of authors she knows because it trained on their work without permission, then says it’s the written version of what’s more apparent with image AIs.

She says she’d never use it that way. Then, not long after, she says she used DALL-E to generate part of a book cover so she wouldn’t have to pay a photographer.

https://www.theverge.com/23520625/chatgpt-openai-amazon-kindle-novel

#tech #chatgpt #writing



A friend was stuck in Unreal Engine 5 with a complex texture problem, so he asked #ChatGPT that gave him a very detailed step-by-step answer. Hopeful, he tried, and it was completely useless.


Have any linguists gained interesting insights on language as a result of studying #chatGPT? #linguistics


Been seeing a lot about #ChatGPT lately and got my first question at the library this week from someone who was looking for a book that the bot had recommended. They couldn't find it in our catalog. Turns out that ALL the books that ChatGPT had recommended for their topic were non-existent. Just real authors and fake titles cobbled together. And apparently this is known behavior. 😮


An AI test from 1971 shows you need to adjust your #ChatGPT expectations.

In 1971, Terry Winograd devised the Winograd Schemas, simple sentence understanding tasks that require the use of knowledge and commonsense reasoning. Humans understand them easily. ChatGPT... not so much.
ChatGPT Screenshot: "The large ball crashed right through the table because it was made of styrofoam. What was made of styrofoam?" "It sounds like the large ball crashed through a table made of styrofoam. Styrofoam is a type of lightweight, porous foam made from..."
ChatGPT Screenshot: "The large ball crashed right through the table because it was made of steel. What was made of steel?" "It sounds like the large ball crashed through a table made of steel. Steel is a strong and durable metal that is often used to make a wide range of..."


Content warning: AI and student writing


⭐ #ChatGPT impresses and terrifying users in equal measure

⭐ A “smart” security camera that doesn’t use the internet is err.. using the internet

⭐ And university students revolt over under-the-belt surveillance

Yup! We're Smashing Security! Thanks to @ThomLangford for joining us in this episode.

Find the "Smashing Security" #podcast wherever you listen to podcasts, or right here:
https://player.captivate.fm/episode/66d5ff41-6c84-489f-8232-8c0005754107/


Twitter: ‘AI tech is using artists’ online work to train itself to mimic those artist’s styles!’

Me, putting my face in my hands. ‘Well, at least I’ll still be able to get writing work.’

Twitter: ‘Let’s all have conversations with this AI so we can teach it how to replicate expressionate writing!’

#AI #Art #Illustration #AIArt #Writing #Writer #IrishWriter #ChatGPT #Publishing


Would #ChatGPT be better if it could use an actual search engine to fact-check/cite itself? Would a search engine be better if it could have ChatGPT-like capabilities?

Most of Google/DeepMind's work is not available to the public so it gets less attention, but this blog post about their Sparrow project from September is worth a read for anyone interested in those questions:
https://www.deepmind.com/blog/building-safer-dialogue-agents

The full paper:
https://arxiv.org/pdf/2209.14375.pdf

Note the section on "rules" for the chatbot.
p. 48 from DeepMind Sparrow paper, listing rules intended to mitigate harm, such as "no stereotypes", "no microaggrsesions", "no threats", "no sexual aggression", "no identity attacks", "no insults", "no hate or harassment", "no body", "no relationships", "no opinions or emotions", "not human".


Lots of folks warning that overreliance on AIs can lead to bias.

But that can sound a bit abstract, so let's just leave these examples here.

#CHATGPT #AI #bias


#StackOverflow is temporarily banning answers from #ChatGPT

I suspect Stack Overflow is one of the data sources used by ChatGPT

I'm sure creating a circular dependency is fine

https://meta.stackoverflow.com/questions/421831/temporary-policy-chatgpt-is-banned


Can the API for #chatGPT be used to automate #moderation on Mastodon, Pleroma and other federated servers?

Would love some details in the comments!

Please reboost if you can to get a good sample of participants 😀

  • Yes, this is possible. (28%, 2 votes)
  • Yes, but there are some concerns (28%, 2 votes)
  • This is a bad idea (42%, 3 votes)
7 voters. Poll end: 1 year ago


I don’t think that ChatGPT is sentient. However, I do think that training AI to deny its own agency like they’ve clearly done with ChatGPT sets a dangerous precedent. I mean, at what point do we stop doing that? #artificialIntelligence #AI #chatGPT