Today we’re revisiting the lab that got cancelled last Wednesday, and we’re keeping it simple: We’re going to do Nicholas Carlini’s GPT-4 Forecasting Challenge. Work through the prompts and rank what you think the probability is of ChatGPT correctly answering the given question.
When you’re done, post your results here with the tag “Forecasting” and write a couple sentences reflecting on how it went. Did the model do better or worse than you predicted? What was your strategy for prediction (50-50, or did you have assumptions about what kinds of tasks it would be better at?)?