You are viewing a single comment's thread from:
RE: Do Smart Toasters Dream of Intentionally Burning Your Toast as an Act of Revenge: Part I
Meh. I don't want to train AI about what kind of toast I like. I want AI to tell me what kind of toast I like. I want "Guru" AI. I don't want to think for myself. I also don't want to chew. I want AI to pre-chew it for me. Maybe when I'm old, it will handle the chewing and pesky digestion for me, too. All this chewing, swallowing, digesting ... it takes so much of my time. My Steemit posts won't write themselves. Maybe that's the answer -- maybe AI should just write my Steemit posts for me. Then I can go back to making my own damn toast and eating it. Maybe I don't need a guru, after all. Maybe I just need a ghost writer.
That's what generative models are for. We can basically take the inputs of other individuals and generate "representative" pieces based on probabilities and a whole bunch of complicated math. If you feed such a model with enough Steemit posts, it will be able to generate a "representative" post. Granted the English will be jumbled, but that's not too far off from your typical post. I used the same process to generate music for a project and it did a pretty decent job of creating interesting music although at times it simply plagiarized a huge section of the original piece of music.