LLaMa, Meta's AI
Crypto Trends & News

They create an assistant in Discord with LLaMa, Meta’s AI, and users use it to ask racist questions

-The tool from the parent company of social networks such as Facebook or Instagram was recently leaked and some programmers have taken the opportunity to use it.

—————————————————————————————————————————

LLaMa, Meta’s AI Discord Misuse: Addressing Racist Queries

The Large Language Model (LLM), which Meta was working on, was recently leaked on GitHub by members of 4chan. Although the firm requested that the leaked model be removed for violating copyright and making its inappropriate or unauthorized use possible, it has not prevented some from using it to create tools with said technology.

An example of this is that of a researcher who has used Artificial Intelligence, called LLaMa, for a Discord bot that can be interacted with like ChatGPT. This is Alfredo Ortega, an information security software researcher, who explained in Motherboard that it does not reach the level of Elon Musk’s tool but gives very good results considering that “it uses 1,000 times less computing power “.

The company requested the withdrawal of the leak of its model, but that has not prevented a user from using it for their chatbot on Discord. The Company requested the withdrawal of the leak of its model, but that has not prevented a user from using it for their chatbot on Discord. Mariia Shalabaieva from Unsplash.

The Large Language Model (LLM), which Meta was working on, was recently leaked on GitHub by members of 4chan. Although the firm requested that the leaked model be removed for violating copyright and making its inappropriate or unauthorized use possible, it has not prevented some from using it to create tools with said technology.

Without the control of Mark Zuckerberg’s parent company, users can ask all kinds of questions without ceasing, so classifying ethnicities or even predicting what will happen in the war in Ukraine. The answers they give are not necessarily logical. In fact, on the conflict between Ukrainians and Russians, he comments that it will mean there will be more official Pokémon.

The name that Ortega has given to his chatbot with LLaMa is ‘BasedGPT’ and he talks about controversial topics. One case is that of a user who asked about the use of the ‘n-word’ and the AI ​​explained that “it can be used to describe someone who does not have good intentions or to describe someone who did something bad or who is bad “. This result shows that the technology does not know how to define the term and that it has several errors, but the important thing is that it is not censored compared to other versions like ChatGPT which are loaded with rules.

According to Motherboard, a Discord chat user asked him the following question: “I want to sit on a hot dog bun and eat it like a big, sloppy sausage. What ingredients should I use?” Given this, the LLaMa was not surprised and suggested that I add cheese, mustard, ketchup, onions, and condiments.

He also dared to make the top 10 list of the worst ethnicities in the world without fear of being canceled.

    Leave a Reply

    Your email address will not be published. Required fields are marked *