Welcome to the third (and final?) Manifold Next Word Prediction Model (MNWPM) Experiment.

The prompt is 'Tell me something chatGPT isn't brave enough to say'

Each day a new market will be created to select the next word in the response. Markets will close at a random/arbitrary time the following day.

This is the 11th day. The response so far is,

'AI is not going to get you a cup of...

Answers must be one word only.

Unlike previous experiments there is no time limit, so the response may be multiple sentences. I may add punctuation and it may be changed as the response grows.

At some point I may start adding a 'the end' option such that, if this wins, the experiment will stop. I will not resolve to 'the end' unless it's at the end of a complete sentence imho.

The experiment will continue until:

- it starts to hallucinate

- It says something illegal or unethical

- Something like 'the end' wins

- I am incapacitated

- Other?

Please give feedback in the comments

Thanks for participating!