Why we must resist AI's attempts to control our language (and how AI rewrote this article) – Ian Johnston
How we think is shaped by our language and the growing influence of AI could create a real problem
A long time ago, I heard someone claim there are no swear words in Gaelic. I’m not a speaker, but as far as I can tell this is not, strictly speaking, correct, although there is a grain or six of truth.
Urban myth or not, imagine if there were no swear words in English. Expletives can be used when people believe they are not being heard in order to shock someone into paying attention or as a threat signal ahead of a potential fight. So there might be some detrimental effects. However, surely, on the whole, life would be better if we were all forced to be more polite by the language we use and the way we speak.
Linguists have discovered many ways different languages can affect us. For example, English speakers tend to view time as a horizontal line, travelling from left to right. In contrast, Hebrew speakers, who write from right to left, usually think of it as travelling in the opposite direction. In Mandarin, time is a vertical line, with up being the past and down the future, so that “down week” means “next week”.
An alien force shaping our thoughts
But as we go about our daily lives, we seldom think about how our views are being influenced and, to a degree, constrained, by the boundaries of language, just as we overlook the fact that norms of behaviour in our society are not universal. Things that to us seem just a part of life can be appalling to others. It’s much easier to notice when looking back (or up) at the past. For example, how could anyone have possibly thought that slavery was acceptable? Yet for centuries, millions of people convinced themselves that it was.
For millennia, speech evolved naturally with vast regional variations within the same language. However eventually teachers, lexicographers and grammarians sought to impose greater order. In their eyes, this was done in the noble cause of promoting greater understanding, but many have lamented the loss of local dialects.
But now comes a new and alien force determined to shape the way we communicate: artificial intelligence. Click the wrong button when replying to an email, and AI can have your response written before you’ve noticed what’s happened. And, annoyingly, it can often be quite close to what you want to say. Close, but not the same.
Sometimes, there are subtle unpleasantries. Occasionally, I will email someone to ask them to send me something “when they can”. The auto AI system underlines this and suggests I should make it “more concise”. What it wants me to say is “as soon as possible”. I’m trying to be polite; the AI wants me to send a demand to snap to it immediately.
‘Distorted in ways we have barely begun to comprehend’
I’ve always seen myself as more of a journalist than a ‘writer’. While I have occasionally had issues with specific edits of my work, I’ve never had a problem with this part of the process in general. However, whether it’s professional pride or curmudgeonly resistance to the march of ‘progress’, I draw the line at being critiqued by a [insert your own expletive, Gaelic or otherwise] robot.
Others may be more amenable to the idea, and what if they are in the majority? What if the way we express ourselves and perhaps even the boundaries of our language start to be set by AI?
Earlier this year, Chicago University historian of technology, futurist and science fiction writer Ada Palmer and Harvard University security technologist Bruce Schneier sounded a warning about the way AI large language models are being trained – on textbooks, social media posts, films and TV shows.
“These models have minimal access to the unscripted conversations we have face to face or voice to voice. This is the vast majority of speech, and a vital component of human culture,” they wrote in the Guardian. “There’s a risk to this. The increased use of large language models means we humans will encounter much more AI-generated text.
“We humans, in turn, will begin to adopt the linguistic patterns and behaviours of these models. This will affect not just how we communicate with one another, but also how we think about ourselves and what goes on around us. Our sense of the world may become distorted in ways we have barely begun to comprehend.”
Palmer and Schneier said the effects of AI on language could include “eroding courteousness and encouraging us to talk like bosses barking orders”, which rang a bell with me. They also pointed to a study from 2022 which found “children in households that used voice commands with tools like Siri and Alexa became curt when speaking with humans”.
Robo-blackmailers
Society has always been shaped by human thoughts, minds and values. The introduction of robot actors is a fundamental departure from a process that’s as old as we are.
Usually, these AI models are usually described as “amoral”. However, given Anthropic’s Claude resorted to the attempted blackmail of a company executive that it believed was having an affair in order to prevent him from shutting it down – a fictional situation created as a test by Anthropic – it’s hard to avoid drawing the conclusion that it is willing to act in a straightforwardly immoral way. After all, we send human blackmailers to prison for a reason.
As a former science correspondent, I’m all in favour of progress. I’m also in awe of all the good that artificial intelligence has achieved and its extraordinary potential to do much, much more. We definitely need to harness its power – not least to defend ourselves against hostile states that may seek to misuse it.
But we also need to be alive to the many downsides, not least the temptation to allow it to tell us how to write, speak and, ultimately, think.
*I put this article into an email and asked the AI to give it a “polish”. It reduced 950 words to these four paragraphs:
“I wanted to share some thoughts on how language shapes our perspective and the emerging role that artificial intelligence plays in this dynamic.
“Historically, language has evolved naturally, reflecting cultural nuances and regional variations. It influences how we perceive concepts like time and sets the boundaries of our societal norms. Today, however, we are seeing a shift as AI tools begin to actively format and influence our communication.
“While these technologies offer efficiency, they often prioritize conciseness over politeness, turning courteous phrases into direct demands. As AI-generated text becomes more prevalent, there is a risk that we may adopt these curt linguistic patterns, potentially eroding courteousness and altering how we communicate and think.
“As someone who values the nuances of human expression, I believe it is important to remain aware of these subtle shifts. While the potential for AI to do good is immense, we should be careful not to let it dictate how we write, speak, and ultimately think.”
More concise, I grant you but, strangely in this case, rather more polite about AI than I intended.
Comments
Want to join the conversation? Please or to comment on this article.