Above all, I think the way she defines her text robots as explores that explore "whatever parts of language that people usually find inhospitable" are very inspiring to me. Personally I have never though of generative texts in this way. I agree that we often want to create robots that can closely assemble humans, not matter if it's a bot that plays chess or reconstructs languages. Thus, embracing the rawness and the awkwardness of the content that was created by a machine and more importantly, find meanings within it can in some sense open up possibilities to new ways of understanding languages, and more specifically, the holes that we had naturally avoided.
Another that stuck with me is the phrase that "gaps could have been there for a reason", as it might have indications of violence or other things that are harmful. I think this is an important point to make. When we make automated machines and let them out in the world, we often consider what they create are out of our control, and it is just what it is (ex. TayTweets). However, I totally agree with the speaker that we, as the creator of the bots, need to take on the responsibility to make sure that we are actively take percussions against those undesirable outputs.