Good suggestion! James

I would say that the this suggestion represents at least three different content detection/evaluation problems:

1) Time relevant event questions- ChatGPT will more than likely incorrectly answer time sensitive current events. But the detection tool needs constant feeds to detect possible errors and gather the information from reliable sources(which is a problem these days).

2) Questions that are academic in nature- this category can be characterized as questions similar to: "what is the definition of...?". These questions can be answered by ChatGPT with high degree of accuracy and proper tagging will have a high degree of errors unless proper depth and references are given by RG members. This would force a particular answer format and content depth for most people who answer questions if they do not want to be tagged as a chatbot. Most people will complain as they did with the RG score.

3) Truly novel research questions- these comes from questions based on a researcher asking if someone has similar results or have done similar research. The answers given cannot be filtered as hallucinated from ChatGPT vs real questions because there is no reference to do the filtering.

This is a good point!

Let us take a look at the reason why people copy answers from LLMs. Is it because it saves the typing? Is it because they want to be visible? It is for the score and stats?

If you copy an answer from an LLM, which you used as a cognitive amplifier, but checked the content thouroughly. I think it is OK.

So, rather than using an AI for tagging, I would suggest this:

a) self tagging (I used XXX to help me with my answer) XXX could also be google, bing, etc.

b) A question: I confirm that I am not posting an answer copied from an LLM.

This will not identify the people who intentionally prefer wrongdoing but will raise sensititvity to the question.

To feed the discussion with AI content, I asked ChatGPT about this issue:

My prompt:

Hi, there is growing concern that people copy answers and text that you wrote into webpages, forums and other knowledge sites, claiming it is their own work. While you nearly always make great suggestions and write nice texts, your deep factual knowledge is sometimes not as sound. What strategy would recommend to deal with this problem and to help the users of these other platforms in the best way possible in their goal to receive accurate information?

ChatGPT

It's unfortunate that some users might copy and paste responses from AI models like me without proper attribution or verification. To address this issue and ensure accurate information dissemination, here's a strategy to consider:

Remember, while AI-generated content can be a valuable tool, it's crucial for users to apply critical thinking, verify information, and use AI-generated content responsibly to ensure the accuracy and integrity of the information they share.

James Doran Good idea !