The Unanticipated Common Ground Between Us and Our AI Counterparts
In the bustling academic environment of my institution, there’s a hot potato being passed around – ChatGPT. As faculty members are embroiled in discussions on academic integrity, those steering the ship are extolling us to “hitch our wagons” to the promising upsides of this “unfamiliar terrain.”
Sure, it’s novel, no arguments there. When a tech journalist can’t persuade a chatbot to stop professing undying affection, we’re wading in uncharted waters. But I propose that the demarcation line between human and machine, particularly in the realm of interaction, is a bit more blurry than many would feel comfortable conceding. And this ambiguity is stirring up quite a hullabaloo around ChatGPT.
Picture this – you’re filling out an online form, and you are asked to verify that you’re not a robot. It’s a no-brainer, right? But then, your email client starts suggesting words to complete your sentences, or your phone attempts to predict your next text. Suddenly, the line becomes blurred. Are you part bot? Are they part human? This raises questions about the very nature of language and our relationship with AI.
The Mirror Effect
ChatGPT, and other AI chatbots, act like a mirror, reflecting human language back at us. They reveal that humans are simultaneously quite original and extremely mundane when it comes to language. This duality, already present in our interactions with each other, is only amplified by our interactions with AI.
Are We Originators or Copycats?
Recently, Noam Chomsky, the renowned linguist, along with his team, suggested that chatbots are trapped in a “prehuman or nonhuman phase of cognitive evolution,” as they can only describe and predict, rather than explain. This echoes Chomsky’s historic recognition of the generative ability of the human language – an ability that seems to elude our AI counterparts.
But there’s a snag in this argument. While humans do have the ability to endlessly produce new strings of language, we often don’t. We’re habitual recyclers, reshaping our speech based on prior encounters, whether conscious or not.
AI and the Human Condition
The prevailing view of human language is that communication is primarily an invention of new phrases from scratch. This assumption, however, is called into question when we observe how AI, like the therapy app Woebot, is trained to interact with humans. Or when ChatGPT is directed to generate lyrics and chords in the style of Colin Meloy from The Decemberists, and the outcome, while mediocre, resonates with the Decemberists’ aura.
This raises some serious questions. Are we, as Chomsky suggests, truly originators of language, or are we more like Bakhtin’s recyclers? Are we part robot? Are the robots part human?
ChatGPT forces us to revisit an age-old conundrum: Just how much of our language is truly ours?
Source: fortune.com