In his concurrance in
Snell v. United Ins. Co. (11th Cir. 2024),
Judge Newsom asked ChatGPT:
Each box represents a segment of text GPT-3.5 could generate. The number on the left indicates the probability of GPT-3.5 selecting that text.
The colored bar shows the likelihood of GPT's responses to Newsom's follow-up question, given the response so far.
For example, there is a 41% probability of an answer starting with: The ordinary meaning of. After that, there is a 79% chance that GPT-3.5 will add the word "landscaping".
- to expand an option: click on box
- to expand all visible options: press the + key.
This website is associated with (citation pending spring 2025 law review submission cycle). If you use or share this tool in your work, please cite that aritcle. Thanks.