When ChatGPT was introduced, it was a big deal. With just a few words, it could write code, poems, or even help make decisions. Other AI models, like MidJourney and DALLE 3, demonstrate remarkable abilities. Just a short prompt and you'll get compelling results.
While Stable Diffusion 1.5 still needed a lot of extras in the prompt such as “only 2 legs, 5 fingers, …”, prompts got shorter. Machines understand our intent better and better:
DALLE 3, prompt: “An illustration of an avocado sitting in a therapist's chair, saying 'I just feel so empty inside' with a pit-sized hole in its center. The therapist, a spoon, scribbles notes.”
Further, frameworks like AutoGPT try to increase the leverage one prompt has by trying to figure out the tasks to do to accomplish the goal on its own.
“Buy my groceries” might soon be possible.
And while AutoGPT has many limitations and, while having flashy demos, can’t yet be relied upon for many tasks, a recent example of raising the “prompt to an impressive result” bar is v0.dev. It showed that we can create a whole user interface with just a few prompts. This makes me wonder: as AI gets better, how much can we achieve with a single prompt?
Think about this:
“Find a cure for all diseases and share it with everyone.”
"Start a successful online business, sell it, and transfer the money to me."
“Build my ideal house and give me the keys.”
These ideas might sound far-fetched now, but they show where AI could go. The big question is, how much can we get done with just one prompt? As AI keeps improving, we might be surprised by the answer. How high will the leverage become?