John Gruber describes his experience using ELIZA, the famous proto-chatbot, from a contemporary perspective. He concludes:

There’s something wrong with people who consider today’s chatbots to be their friends or companions. But there’s something really wrong with people who considered ELIZA a useful confidant. Like suffered-a-permanent-head-injury wrong.

This past Monday, as I lay in a groggy late-morning fog after a much-extended wee-hours flight, I received an unexpected text. It was from a person with whom I’d had a falling out, some months back. They seemed to want to patch things up, and asked me to explain myself about certain specifics that continued to bother them.

The message energized me, and not unpleasantly. I took it as a welcome challenge, and spent the next several hours composing a lengthy reply, doing my best to understand their perspective, and how my past actions must have caused painful friction which led to misunderstanding, and then resentment. I apologized for my errors in communication, pledged to improve, and stated a desire to resume our friendship.

My interlocutor gave my reply due consideration, then said that I had confirmed their earlier profile of me as a delusional liar with all the social worth of a rotting tangerine. And so ended that particular conversation.

But still: it felt nice to be asked. In particular, I welcomed the unexpected prompt to describe a problem I had—even if I wasn’t aware any such problem prior to being asked. This in turn encouraged me to engage in enough novel self-analysis to write out a thoughtful reply, digging up new truths about myself in the process.

So that’s me thinking that the principle holds even when the question is asked in bad faith, or if the entity asking the question is itself a selfless void.

“Tangerines” by neil conway is licensed under CC BY 2.0 .