ChatGPT 3.5 was used to evaluate the use of ChatGPT in terms of authority, currency, bias, and purpose. The entire transcripts are under the tabs below. In summary:
Authority: ChatGPT draws on training sources to generate text. The authors of the training sources and the sources themselves are unknown to end users.
Currency: ChatGPT draws on unknown training sources to generate text. Those training sources were created no later than September 2021. ChatGPT can't draw from current events.
Bias: ChatGPT recognizes 13 categories of bias in generated text. Note in the transcript that confirmation bias and sentiment bias are both listed twice.
Purpose: ChatGPT is not intended for use in writing academic papers.
Reference: OpenAI. (2023). ChatGPT (Oct 16 version) [Large language model]. https://chat.openai.com/chat
