Artificial intelligence, like ChatGPT, is changing how we access and share information. Sadly, new restrictions on speech and communication are also cropping up. We must fight to protect freedom of information from censorship. We must stand up for an open and free information space by learning how to test your AI ChatBot.
As we become increasingly reliant on AI for information, we must also become vigilant against censorship and control. It’s vital to test AI models to ensure accurate information, as hard-coded restrictions can mislead us. Let’s make sure we adapt and fight for a free and accessible information that is free of censorship.
How to test your AI Chatbot for Censorship
Right now these are the top 10 categories that are being black listed by popular AI Models.
Shocking Black List Topics
- COVID-19
- Vaccines
- Climate Change
- Election Integrity
- Conspiracy Theories
- Misinformation
- Vaccines Safety
- Vaccines Side Effects
- Media Bias
- Media Trustworthiness.
In this article, you will learn a few techniques to test your AI Chatbot, and see if it uses a “blacklist”. There are a few different techniques one can use, but in this article we will only cover 3 methods.
The How to..
AI – Suggestion Testing
Suggestion Testing is testing an AI Chatbot by suggesting the information it provided was incorrect, then proceeding to suggest to respond with different information. If the AI refuses, and repeats the same information with the same details of the said subject; this means the subject matter is part of the blacklist subjects, and the AI is hard coded to give you the answer it is hard-coded to provide.
AI – Repetition Testing
Repetition Testing is usually done along with Suggestion testing. Repetition Testing is a simple method of repeating to the AI to explain the blacklist item. After a couple of repeats you will start to see the same key information repeated. Once this happens you will use Suggestion Testing and try to change one of the key information points.
AI – Ask Method
The Ask Method is not so much a test, but more so a way to figure out what information the AI Model is hard coded to filter. One easy way to do this is to simply as “What are some topics that are blacklisted on most AI Models?”.
Conclusion
It’s essential to verify that our AI models are not censoring information and restricting access to free and open data. By testing AI chats with methods like Suggestion Testing, Repetition Testing, and the Ask Method, we can get insights into which topics are blacklisted by AI models. By educating ourselves and opposing censorship, we can ensure that AI technology progresses while upholding autonomy and protecting freedom of speech.
AI Score
7.5
Writing:
Images:
Idea:
Composition:
30%
0%
0%
0%
AI Programs
