2020 Review (GPT-3)| AI as a poet, novelist, and dramaturg

My GPT-3 Year 2020

Call me snobbish, call me a freak, call me n00b. There are so many Natural Language Processings models out there. Many of them are presented by fancy graphs, morphological analytics. Some models can generate a sentence in several languages at the same time. Another can rapidly translate simple messages. Some are the most efficient. Others can correctly Question&Answers games. But that’s not enough for me.

Many language models are already developed for business use, to fulfill a chatbot function, to understand a customer.

But I want a language model that is can write stories for me. Original, funny, absurd, surreal stories. Large texts. Unsupervised. This irrational urge of mine will spook out many ML researchers. If the text doesn’t make sense, why long for this? We even cannot benchmark it. For which purpose? For which approach?

My Approach

Yes, it’s about my approach, my dream, my vision I am following already since Google Deep Dream, consequently applying in Replika and StyleGAN.

I seek for a dialog with the machine about its perception of art, human culture, reality. Not what we as humans are expecting to…