Using ChatGPT and Other AI-based Technologies for Lawyers

1. Do not input any confidential information into these programs. OpenAI is very clear that any data inserted into the program is saved in an effort to train and improve their models. Inputting client data could result in ethics violations such as a breach of confidentiality.

2. Please double check any information generated by ChatGPT for accuracy. For example, when dealing with legal information, ChatGPT often will generate fictional case citations. Lawyers have already been sanctioned for submitting these fake citations (see The ChatGPT Lawyer Explains Himself).

3. Do not rely on ChatGPT for developing arguments with deep context. ChatGPT is excellent at mimicking past content that its models have trained on; however, it is less effective at understanding subtext or developing subtleties within an argument. This trait makes it much more effective for generating standard corporate documents rather than bespoke, dense litigation arguments.

4. Please ensure the output data from ChatGPT is updated (and based on current law). ChatGPT has only been trained with data up until the end of 2021. In law especially, new cases can change the status of a legal issue very quickly.