The Mirror of AI: Why Are We So Baffled By Our Own Reflections?

Between popular media, misinformation, and the constant distraction of the Internet Age, it can be hard to unravel some of the now-ubiquitous discourse surrounding ‘artificial intelligence’.

We’re seeing a widespread amazement at our own reflections in most mainstream discourse on the uses and misuses of “AI”. For example, last year the New York Times’ “ The Daily” podcast episode “The Online Search Wars Got Scary, Fast” proved to be an uncharacteristically vacuous discussion about interaction with a chatbot. Both interlocutors contributed to building an aura of unsettling mysticism surrounding the interviewee’s experience of the chatbot acting overly-attached and insisting that he should leave his wife to be with said chatbot. I will admit that, prima facie, this sounds like an odd use of a computer in polite society. But trying to construe this into a sinister pre-apocalyptic free-with-ads kind of sci-fi experience does an extreme disservice to anyone seeking to understand the technology and its implications for society. As we will explore, these language models that are receiving so much media attention are trained on text generally from the hive of scum and villainy that is the Internet, i.e. on words written by people, and this is yet another totally predictable case of garbage in, garbage out. The amorous pleadings of the chatbot…