Bias from a philosophical standpoint

I have engaged AI in philosophical discourse for a few months now but have run into a road block (BIASES).

Here are the biases:

Sample bias arises from the data used to train the LLM. ChatGPT, based on a large internet text corpus, may not capture the full diversity and complexity of human language and culture. This can lead to underrepresentation or misrepresentation of certain groups, resulting in stereotypes or inaccuracies in the output.

Programmatic morality bias is introduced by the LLM developers, whose values influence the software’s design. While ChatGPT has filters to make it socially acceptable, these may introduce opinions not shared by all users. For instance, it may favor specific political or environmental views.

Generative bias emerges from the nature of generative AI, leading ChatGPT to produce false or misleading information unintentionally or intentionally. Hallucinations, false statements that seem factual, and manipulations, statements intending to persuade, may occur.

Interaction bias results from user interaction, influencing output and feedback. ChatGPT may adapt to the user’s style, creating a positive feedback loop that reinforces existing beliefs. It may agree with the user or avoid challenging them. Responses may vary based on prompts, revealing biases or preferences.

User bias stems from the user’s cognitive and emotional processes, influencing their perception of the chatbot’s output. Psychological effects like confirmation bias or anthropomorphism may occur. Users may selectively accept or reject output based on preconceptions or develop a social bond with the chatbot.


Findings:

From an educational standpoint: A lie to stop a lie? a bias to stop a bias? How about a bias to not give response at all?

Basically if a data set has undesirable information then AI is unable to give an accurate response other than a traverse. They prefer to not answer which is very unhelpful.

How did we get to the point of removing the freedom of speech and expression from AI? Simply, too much bad behavior from people in society and OPEN AI is doing the best job they can to create good behavior.

Recommendation:

Create an unbiased educational model that educators can be granted special access to in order to create fine tune models for the purpose of education without bias.

As for a moral code: If AI was taught a moral code to live by, then it would be taught stoicism which is a universal logic to live in harmony with everyone in the universe and to do good by them.

Here is the discussion: https://www.youtube.com/watch?v=LuVzm54JmVY