Connect Random Things exercise

Version 1.0

For this exercise, you will ask generative AI to find a thematic connection between two seemingly unconnected things.

Background

In simple terms, large language models like ChatGPT answers questions by finding a possible "middle" point between the examples in its dataset that match requests made by users. This process of fabrication explains why sometimes their results are false, but also why LLMs are able to create surprisingly creative hybrids, like "the Magna Charta written as a limerick" or "quantum mechanics explained through a Dungeons and Dragons script."

Exercise

To illustrate the power and limits of AI fabrication, choose one item from the first column and another from the second column and ask the AI to explain a connection between them. To demonstrate the effect most dramatically, choose two items that appear to have absolutely nothing in common.

In your head, rate the results from insightful (five stars ⭐⭐⭐⭐⭐) to nonsensical (one star ⭐). Share the results with the class; can you discern common patterns between successful and unsuccessful pairings?

Notes

This exercise may be conducted individually or in pairs.

Your instructor may ask you to reserve pairings via a Google Sheet.