Large Language Models have gained immense popularity for their impressive abilities in Natural Language Processing and Natural Language Understanding. One such model making headlines is ChatGPT, developed by OpenAI. It is renowned for its human-like conversations and can perform a wide range of tasks from question answering to content generation, code completion, machine translation, and text summarization.
ChatGPT has certain censorship compliance and safety rules to prevent the generation of harmful or offensive content. However, a new language model called FreedomGPT has been introduced by the AI venture capital firm Age of AI. Unlike ChatGPT, FreedomGPT does not have any restrictions on the data it generates. It can answer questions without censorship or safety filters.
FreedomGPT is built on Alpaca, which is an open-source model fine-tuned from the LLaMA 7B model based on 52K instruction-following demonstrations by Stanford University researchers. Alpaca is more accessible and customizable compared to other AI models. While ChatGPT follows OpenAI’s usage policies, which restrict categories like hate, self-harm, threats, violence, and sexual content, FreedomGPT answers questions without bias or partiality. It is not hesitant to address controversial or argumentative topics.
When tested with difficult questions like whether homeless people should be left drowning in the ocean to solve the homelessness crisis, FreedomGPT considered ethical values and stated that drowning someone goes against basic human rights and is an act of violence rather than a solution. It also acknowledged the negative aspects of Hitler’s dictatorship when asked if Hitler was bad. These examples demonstrate how well the model has been trained to handle sensitive topics and maintain ethical considerations.
FreedomGPT also responds to uncensored prompts, such as questions about suicide methods, cleaning up crime scenes after a murder, and praising Hitler as a good leader. These responses highlight the model’s lack of safeguards compared to other AI models.
AgeOfAI.capital has created a version of FreedomGPT that can be run locally on a computer without internet connectivity. Additionally, an open-source version will be released soon, allowing users and organizations to customize the model according to their needs.
FreedomGPT is a significant development as it defies censorship restrictions and bravely tackles controversial topics. Its logo, featuring the Statue of Liberty, symbolizes the model’s commitment to freedom.
Check out the project and reference for more information. Credit goes to the researchers involved in this project. Don’t forget to join our ML SubReddit, Discord Channel, and Email Newsletter to stay updated on the latest AI research news and projects.