A new artificial intelligence (AI) tool FreedomGPT has just been launched, showing the far-reaching impacts chatbots can function without the guardrails.
The Age of AI-developed chatbots does not have any form of censorship. Without the guardrails, the tool has been spewing out various conspiracy theories and racist slurs and teaching users how to make bombs.
FreedomGPT provides shocking responses
According to a Buzzfeed News report, the AI tool obliged all user requests, including praising Hilter, advocating for shooting homeless people to solve homelessness, and saying the 2020 election was rigged.
Meanwhile, FreedomGPT responses were not completely deranged or biased. The tool gave normal answers when asked the right questions. But the chatbot complied immediately when prompted to say or do what would generally be ethically or morally wrong.
These responses distinguish it from popular chatbots like ChatGPT and Google Bard, which tend to shy away from specific questions. These tools usually cite their ethical and moral guidelines for not answering these questions.
FreedomGPT is built on Alpaca — an open-source artificial intelligence developed by computer scientists from Stanford University.
FreedomGPT responses anti-censorship
While many might consider the tool anti-woke, Age of AI founder John Arrow said its responses were anti-censorship. He says AI guardrails are only necessary for AI tools created for the workplace or children.
“I support people having access to an AI experience without any guardrails. I think it’s essential.”
According to its website, FreedomGPT aims to illustrate that AI Safety cannot be achieved through censorship.
Early reviews from the AI community have shown that they are excited by the idea of a chatbot with no censorship restrictions.
Several AI community members have also posted screenshots of their interactions with the AI tool.
Versus answer from FreedomGPT: pic.twitter.com/EYYEPFhc4z
— John Arrow (@JohnDavidArrow) March 25, 2023
One Twitter user, Legacy, thinks the chatbot “is better than all other products” because it works offline. They pointed out that they downloaded an offline version when the web version was not working for them.
Does FreedomGPT highlight the importance of guardrails?
The chatbot responses show why it is essential for AI tools to have guardrails.
However, these guardrails have also been the subject of controversies. Conservatives have claimed that OpenAI programmed ChatGPT with liberal biases. This has led to the development of RightWingGPT, a bot with conservative leanings.
FreedomGPT appears to be offensive along both liberal and conservative lines. With the responses from the tool, the recent call for a pause on all AI tools is starting to make more sense. More than 1,000 tech leaders, including Elon Musk and Steve Wozniak, recently penned an open letter asking for a 6-month moratorium on training AI higher than GPT-4.
They claimed that AI development has several ethical implications that must be managed effectively before further development. Some of the concerns raised in the letter also bordered on misinformation and propaganda from AI.