Does anyone else think “Prompt Engineering” comes across like “learning how to how work around the ways that LLMs actually aren't intelligent”?
Mind you, I _do_ see value in that skill set. It just feels weird to declare a new form of engineering around it.
@LouisIngenthron @msilpala interesting… “prompt engineering” is the effort to try to get the AI to behave more like a computer!
That’s hilarious 😂
@narthur @LouisIngenthron I find the LLM/AI hype machine to be remarkable. When else has a new class of product arrived and sparked BOTH of these responses:
"This is amazing, and will soon be making every kind of job obsolete.”
"It takes work and practice before you can get it to produce the intended results—and, by the way, sometimes it's just plain wrong.”
I do believe this is amazing stuff; it will do wonderful and terrible things. For now it is producing WAY more enthusiasm than value.
@msilpala Not really. If anything, it's the opposite. Prompt engineering is an attempt to make the free-wheeling responses of AI conform to a set pattern so they can be used by other processes.
While you can always do that by having a conversation with the AI and slowly leading it, bit by bit, to where you want to go, the prompt engineer works to get there, consistently, in one request.
Therefore, so long as "social engineering" remains a valid term, I see no problem with this term as well.