I’ve been reading The Most Human Human by Brian Christian. Christian is an American writer; the book charts his experience, research, and ideas about going up against computers in a Turing test. In the Loebner Prize contest, humans and AI programs compete to be found human. Judges conduct text conversations with the competitors, and award points. AI programmers covet the title of ‘The most human computer’. The title of Christian’s book comes from the award given to the human competitor that gets the most points from judges.
In the chapter I’ve just been reading, Christian talks about a key difference between the best of human conversation and how AIs ‘converse’. For meaningful human conversation, he says, you need a stable point of view, a unifying vision, a single style, and a memory of the current conversation and any previous conversations you’ve had with the same person. Furthermore, each conversation with the same person builds on the emotional connection made in previous conversations.
AIs, by contrast, excel at conversations in which each reply depends only on the previous remark, without any need to remain consistent and without any knowledge of the history of the conversation or the relationship. Christian follows programmer Richard Wallace in calling such conversations ‘state-less’. That is, remarks don’t depend on any pre-existing state.
The type of human conversation that most closely resembles AI conversation, Christian claims, is verbal abuse. Christian offers a typical exchange:
‘Oh, there you go right in with that tone of yours!’
‘Great, let’s just dodge the issue and talk about my tone instead! You’re so defensive!’
‘You’re the one being defensive! This is like the time you x.’
‘For the millionth time, I didn’t remotely x. You’re the one who…’
Each remark after the first is only about the previous remark.
Doesn’t this sound like some of the comment streams we’ve all known and hated? Christian goes on to say that such interactions are reflex reactions to the last comment rather than:
either the actual issue at hand or the person I’m talking to. All of a sudden the absurdity and ridiculousness of this kind of escalation become quantitatively clear, and, contemptuously unwilling to act like a bot, I steer myself toward a more ‘stateful’ response: better living through science.