Submission Statement
Lots of people in the world of SEO, marketing, and copywriting are excited about the possibility of creating vast quantities of AI-generated text. They have a problem. Human waking hours are finite, and many of us may be near our upper limit for absorbing new content.
OP examines the other side of this. How new AI’s advances allow us to examine text. It seems obvious to me this will have more profound effects than the ability to generate text. Consider one aspect of this.
AI should allow us to analyze the logic in politicians’ speeches in real-time. There are over a hundred logical fallacies, and they are a standard part of political debate. So much so, if you took all the logical fallacies out of political debate - what would you have left? Soon people may have the ability to easily find out.
AI should allow us to analyze the logic in politicians’ speeches in real-time. There are over a hundred logical fallacies, and they are a standard part of political debate.
This seems interesting. For starters, I wonder if a model could be trained to spot logical errors in mathematical proofs. That in itself could be a useful tool for mathematicians, but it seems like an error detection model could also be used to improve existing text generation models through adversarial training.
