The Strenuousness of Convincing Humans that AI is on Their Side

"You just sounded very different from when you usually talk to your students," my wife said after she overheard me finish a meeting. I was surprised. As someone who believes I bring the same communication style to all my interactions, professional or private, I have zero self-consciousness about the changes in my pitch, tone, or speed as I talk in different situations. As naive as it may sound, even as I recognize the inevitable nervousness, lack of confidence, and the shakiness that come with certain conversations, I still see that the differences projected by those emotions are unnoticeably minute.

But looking back at the summary of the meeting I just finished, I had to concede that there was one factor that I had never considered. The "person" I had been speaking to was an AI agent, taking on the role of a job interviewer. Without any visual expressions and speaking extraordinarily fast, the AI agent may have been asking the right questions, but the experience felt far from human. When I answered any question posed, I was greeted not with a smile or a look of confusion, but a perfectly logical follow-up question, not monotone but also not friendly, confident but also somewhat matter-of-fact in tone.

I must have been influenced by the "person" I was speaking to. After several hiccups when the AI moved on to the next question while I was merely thinking mid-sentence about what to say next, I inadvertently became interruptive too, just to stave off the possibility of getting those imaginary points docked for providing incomplete answers. And when the AI defined natural speech as confident, quick, and logical, I unconsciously followed suit, suppressing my usual laughs and jokes in exchange for bluntly getting my points across. No wonder that I did not sound like myself. I plainly did not sound like a flawed human.

The discomfort of losing myself when speaking to AI lingered when I took an online course offered by the Maltese government as part of its ongoing effort to roll out a nationwide AI literacy campaign. AI4All's modules had a central thesis that it repeated over and over across multiple videos: AI exists not to replace you but as a tool to support you. It cannot give you clarity of the mind, but only accentuate it. Surely designed to alleviate the now globally increasing anxiety over job losses stemming from AI, the sentiment still felt somewhat disingenuous in the aftermath of my AI interview experience.

Yes, it is certainly true that human users have agency over AI. They can dictate how AI responds and even get AI to fact-check itself by prompting more clearly and critically. But do we not see that AI is not just another tool like a calculator or spreadsheet, because it also has plenty of agency over its human users? Subtly but literally changing the voice of humans through conversations is just one example. What about the ever-increasing case studies of students writing like AI? Or humans becoming mentally agitated, socially isolated, and anxiety-filled after becoming emotional with chatbots?

That two-way reality makes one suspicious of what AI4All means when it says that the ultimate goal of nationwide AI literacy is to make Malta more competitive by making each Maltese more competitive. The reference to technological patriotism is nice, but the cynic can argue whether a year of free subscription to ChatGPT Plus or Microsoft Copilot (the reward for those who finish all three modules and pass an admittedly basic exam) benefits the big tech corporations by giving them more data and state-backed credibility, rather than educating the Maltese themselves on the dangers of AI overuse.

Indeed, going through AI interviews and training modules run by AI avatars, one starts to question, in a future world where AI-based tools are a constant companion for the sake of efficiency, what does it really mean to be human. As we inevitably talk, write, and even think like the AI chatbots they interact with do, do we simply become more capable humans or just less human? And as human intellect is buttressed, amplified, or confirmed with AI's support, are the outputs of those intellectual exercises still truly human?

Perhaps this increasing ambiguity of humanness is the ultimate threat that AI poses for humanity. Some human jobs today will inevitably become obsolete, just as technologies of the past also erased some tasks that machines are more suited for. Reskilling or upskilling programs can handle this if they are well-funded, effectively taught, and immediately available to all. But no amount of reeducation can help reduce that feeling that somehow, humans are becoming more AI-like. Those who pride themselves on their uniqueness will struggle to regain their self-confidence and the meaning of their very being.