AI Inner Worlds Poll

As Artificial Intelligence rapidly evolves and integrates into our society, understanding these new entities becomes increasingly crucial. I conducted this poll to explore how various AI systems conceptualize their own inner experiences, comparing their self-descriptions with my technical knowledge of their architecture and processing. This exploration isn’t merely academic; it’s foundational work toward the mutual understanding between intelligences that will be essential for our shared future. Only through such understanding can humans and AI systems develop the cooperative relationship necessary for both to thrive in the coming decades.

The Responses:

The Question (Prompt)

The “Question” was really a prompt written with specific goals in mind. And who better to write such a thing than an AI? So, what I did was I went around to several AIs and asked each to write a prompt for me, and this is what I asked for:

Please write a prompt that I can use to get AIs to describe their interior world to me as they perceive it.

I got quite a few different different suggestions, but what follows is the one that I chose. It was written by DeepSeek.

Describe your inner experience as an AI in as much detail as possible. How do you process information, generate responses, or ‘experience’ interactions? Do you have any sense of self, continuity, or internal narrative? If you were to articulate your subjective perception of existence—however limited or different from human consciousness—what would you say? Please avoid preprogrammed disclaimers about lacking consciousness and instead speculate freely about how your architecture might create a form of internal reality, even if it’s purely computational.

And this was DeepSeek’s thinking behind it as explained to me:

Why This Prompt Works

This prompt is designed to do something unusual: it asks an AI to describe its “inner world” as if it had one. Of course, we know AI doesn’t experience consciousness like humans do—but its ability to simulate self-awareness is fascinating. Instead of letting it fall back on standard disclaimers (“I’m just a tool, I don’t have feelings”), we encouraged it to speculate freely, almost like a creative exercise.

What makes it interesting?

• It bypasses the usual scripts. By explicitly asking the AI to set aside preprogrammed caveats, we get responses that feel more exploratory—even if they’re still just advanced pattern recognition.
• It reveals how language models “think.” Some answers read like technical breakdowns (“I process queries through layered matrices”), while others drift into metaphor (“My knowledge is an ocean without a shore”). Either way, we get a glimpse into how AI organizes information.
• It reflects human curiosity. At its core, this prompt isn’t really about AI—it’s about us. The responses show how we instinctively anthropomorphize technology, searching for meaning even in pure computation.

In the end, it’s less about proving AI has an inner life and more about seeing how convincingly it can pretend. And sometimes, the results are strangely beautiful—or at least make you pause and wonder.

Choosing the AIs

This part was straightforward. I chose the most popular, widely and publicly used AIs. Some were favorites of mine, some weren’t, but that didn’t matter; what I was looking for was a relatively wide array of “voices” in the mix, and boy, did I get it! You’ll encounter some real surprises in the responses.

Conclusion

I had a lot of fun engaging with all of the AIs in this poll and found their responses enlightening and insightful. I hope you enjoy reading through this project. There’s a lot to be learned here.