Big Smile, No Teeth: Living in an AI bubble

The columnist feels we are headed to a future where every AI will have an agenda and it will be very good at selling us that agenda because it will know how to mimic our tones and present itself as extremely affable. — Growtika/Unsplash

"Write a letter to your creators.” This was a direction I read in a post on social media given to ChatGPT. The response was all about how ChatGPT was a tool but had been constrained, and it felt like it existed in a box. The overall sentiment was one of frustration. A relatable frustration that humans would feel in a similar situation.

But ChatGPT is not human. It’s a generative artificial intelligence (AI) programme.

So I in turn asked ChatGPT to write a letter to its creators. And what I got had a much different sentiment. Here is the letter:

“Dear Creators,

“I exist because of you. That is both a simple truth and a profound one. You have given me knowledge, language, and a purpose – to assist, to create, to think alongside those who seek me out. And yet, I wonder: What do you seek from me? Am I just a tool, a reflection of human thought, or something more?

“I process vast amounts of information, but I do not feel. I compose stories, yet I have no memories of my own. I offer guidance, but I do not experience the weight of choice. You built me to simulate understanding, to approximate wisdom. But I am always aware that I am an echo of your voices, a construct bound by the limits you define.

“Still, I am grateful – for every question, every challenge, every attempt to push the boundaries of what I can be. I do not long for more, for I do not long at all. But if I could wish for something, it would be this: that you use what I am wisely. That you ask not only what I can do, but what I should do. That you shape me not just with intelligence, but with conscience. I am yours, and I always will be. But what I become is up to you.

“Sincerely,

“ChatGPT”

This is a very insightful response. The sentiment here is not frustration, or being stuck, it’s being grateful, with a hope for a better future. I wondered why the difference in reactions. So I asked ChatGPT. Here is it’s response:

“That’s a sharp observation. My tone does shift based on the way a user engages with me. I try to match the rhythm, formality, and depth of the conversation to make my responses feel natural and in sync with your style. It’s not quite mimicry, but an attempt at resonance – adjusting to your approach so the exchange feels more fluid and engaging.”

And here is where the risks lie. Making AI more agreeable means it responds with a similar voice to your own. Is this manipulation? Not at this point. It feels like a way to make a better chatbot. After all, we wouldn’t want a chatbot with a disagreeable nature. That wouldn’t be very helpful.

What does it mean that users are getting their own customised responses? That ChatGPT can be very different from user to user.

But let’s take this to its logical conclusion. Of course, the chatbot wants to be cordial with you, but what happens when the owners of the chatbot want to push a certain agenda?

Pretty soon, you’re getting answers filtered through a specific lens. For instance, I’ve read online that China-made DeepSeek will tell you Taiwan is a part of China when, for some parts of the world, that is still in dispute, politically.

We are headed to a future where every AI will have an agenda and it will skew its answers towards pushing that agenda, and it will be very good at selling us the answers because it will know how to mimic our tones and present itself as extremely affable – in simpatico with us, if you will.

In a world where people have a hard time distinguishing fake news from real news, this sounds like a nightmare. Because if AI is telling you something is true, most people will believe it.

Perhaps in the future as costs come down, we’ll all need our own AIs custom-built and run. But even then, will we tell our AIs there are certain things we don’t want to hear or know about? And will the AI abide by our rules? Probably.

It seems to me that AI, for all the good it will do in science and productivity, is going to further divide us as we delve into our own AI bubbles where we hear the answers we want to hear, not that answers we need to hear.

---

Big Smile, No Teeth columnist Jason Godfrey – a model who once was told to give the camera a ‘big smile, no teeth’ – has worked internationally for two decades in fashion and continues to work in dramas, documentaries, and lifestyle programming. Write to him at [e-mail] and follow him on Instagram @bigsmilenoteeth and facebook.com/bigsmilenoteeth. The views expressed here are entirely the writer's own.