Gambling Warning For Generative AI #

Generative AI and gambling have random dopamine hits in common, the most powerful addictive hook there is.

Some times the simplest strategies is the one which wins, other times is the most complicated one. What makes it more interesting is that the same strategy used twice might not have the same result. You are always “one more round” away from the win.

This is how the addiction begins. What if I should have generated 10 more images. What if one more code refactoring would have finished the feature implementation? What if the next article rewrite would have been popular?

Due to the fact that the results are predictable, the sunk cost fallacy kicks in even harder. Since there is no specific parameter that you can tweak and see predictable, reproducible changes if you move it along a scale, you don’t even know what is wrong. Is it the prompts? Is it the seed? Is it how the problem is formulated? Is it the right solution but I’m unable to see?

People use generative AI for what it makes them feel not for what it does.