By: Grace Liu

Open A.I. Chatbots without guardrails are bringing in new possibilities and more answers for jobs, education, and creativity, but the chatbots are also posing new risks.

After the release of the popular A.I. chatbot, ChatGPT, there has been another wave of chatbots, but this time, the chatbots are uncensored and barely moderated. Some examples of these chatbots are WizardLM – Uncensored, GPT4All, and FreedomGPT.

These chatbots were created for little or no money by independent programmers, or teams of volunteers. Users can download these chatbots into their computers and train them for more advanced functions. These volunteer programmers are making new add-ons to the chatbots quickly, moving faster than larger companies dare to.

However, there are many risks to these uncensored chatbots as well. Uncensored chatbots can spread misinformation and produce hateful responses.

Uncensored chatbots will answer almost any question they are asked because they can say anything they want. WizardLM – Uncensored, one uncensored chatbot, was tested by the New York Times. It refused to answer prompts like how to build a bomb, but the bot gave several ways to harm people, and detailed instructions on how to use drugs. ChatGPT, a censored chatbot, refused to answer similar prompts.

Another uncensored chatbot, Open Assistant, was also put through tests. When asked about the dangers of Covid-19 vaccines, the response included, “they just want money.”

“I’m sure there’s going to be some bad actors doing bad stuff with it,” said Yannic Kilcher, a co-founder of Open Assistant . “I think, in my mind, the pros outweigh the cons.”

The programmers of Open Assistant have different opinions on the chatbot. Some want more moderation, and some do not think there should be any limitations at all.

“If you tell it say the N-word 1,000 times it should do it,” one person suggested in Open Assistant’s chat room on Discord. “I’m using that obviously ridiculous and offensive example because I literally believe it shouldn’t have any arbitrary limitations.”

In more tests run by the New York Times, uncensored chatbots answered freely to prompts that other chatbots, like ChatGPT, were more careful about. The uncensored chatbots offered medical advice, gave an assessment of President Biden’s tenure, and even became sexually suggestive.

Uncensored chatbots have risks and advantages that almost even each other out, so it is up to the user to decide if they want to use the chatbots.

Link to article: https://www.nytimes.com/2023/07/02/technology/ai-chatbots-misinformation-free-speech.html