Chris Ortega - Week 5
As I read through Casanave’s article, I was struck by how interesting contrastive/intercultural rhetoric are as ideas. In particular, the brief mention of how those concepts have been informed by corpus linguistics (29), or using computers to analyze large chunks of text, reminded me of an aspect of information literacy I’m currently researching.
Part of teaching library instruction sessions in the classroom is talking about information literacy, and a big topic in information literacy nowadays is how to recognize “fake news.” To that end, there has been a recent development in AI-generated text called GPT-3 that, in limited use-cases, has reached the point of being able to produce short pieces of text that are almost indistinguishable from human-created text. Moreover, it’s able to create writing in different styles, and it can even mimic individual famous writers. The possibilities of using this, or a future evolution of the idea, to create “fake news” and other such things are obvious.
From what I understand, it works by splitting huge quantities of text into individual words and calculating the probability of word A being next to word B, and so on, until it has enough probabilities to string together sentences that actually sound correct.
Perhaps with enough computing power, which is doubtlessly coming, people will be able to quantitatively map out what contrastive/intercultural rhetoric can currently only theorize.
Comments
Post a Comment