Discussing with people who no longer exist is one of the most profound challenges humanity could take on. And I’m not trying to evoke an emotional reaction or touch on a sensitive topic. Just think of a person you’ve never met but know a lot about—perhaps from another era.
Simply the possibility of having even a single interaction with such a person feels beyond common sense, beyond physics, and challenges many paradigms that history has taught us.
The very thought of exchanging ideas with someone from another era seems incomprehensible. For most of history, it was regarded as pure magic—a notion belonging to myths and fantasy. But today, AI is bringing us closer to that reality.
On a completely different side of the story, there’s the Turing Test—a concept that defines when a machine can converse so naturally that it becomes indistinguishable from a real human.
Turing, along with the technologists who followed, laid the foundation for machine intelligence, which has evolved to the point where AI can transform raw data into knowledge. After decades of research in artificial intelligence (which is just a fraction of human history), we’ve reached a point where machines can predict the next word in a sentence by being trained on terabytes of text.
Today, machines can process and generate human-like language patterns.
We are in the era of LLMs. What a time to be alive!
Who would have thought that “simply” learning a language could make a machine appear so human?
And as if that weren’t fascinating enough, this raises even deeper philosophical questions:
- Who are we, really?
- How much of our identity is tied to our knowledge?
Because if we gather enough knowledge about someone—especially if they were a prolific writer or speaker—it essentially means we’ve created a snapshot of their mind. Or perhaps, something even better, since the human brain doesn’t store every memory in raw format.
With so many barriers removed by LLMs, we can now transform one’s knowledge into something remarkably close to the person itself. Or at the very least, we can get a glimpse of an experience that was once unimaginable—synthesize a historical figure’s documented knowledge into a conversational experience that closely resembles their style and reasoning.
For me, this isn’t just a curiosity—it’s a breathtaking pursuit, a fascination that drives me to absorb knowledge and use any tools that bring me closer to achieving this.
Sometimes, despite my deep love for technology, I feel that it’s not technology itself that fascinates me the most—but rather what we can achieve with it. Technology is an extraordinary tool, and software in particular empowers us to accomplish things that were undeniably impossible in the past. And that, in turn, opens new philosophical questions we had never even considered before.
But in the end, all the education, practice, and experimentation feel like just vehicles—guiding toward bigger aspirations. One of them? Experiencing a conversation with someone from the past.
This journey started for me a very long time ago, but at least I can pinpoint a moment seven years ago when I decided which historical figure I would want to meet.
My choice: Siddhartha Gautama. Not as a deity, but as a historical figure.
This choice had no religious meaning—I simply had to pick one person from all of human history to answer a question that popped up on Facebook. And I chose him because he was ahead of his time, a pioneer of content creation, and the only person in history to build his own philosophy of life from the ground up.
This idea has driven my research and development for years, leading me to experiment with various AI techniques to achieve it. I’ve even asked myself: What questions would I actually ask him? In which language?!
And now, it’s not even that complicated or expensive to do.
To bring this idea to life, I fine-tuned an LLM using first-hand writings attributed to Siddhartha Gautama, carefully curating data to ensure that the AI’s expressiveness and reasoning align with historical texts. I incorporated Chain-of-Thought (CoT) reasoning for deeper contextual awareness, combined two LLMs to enhance response accuracy, and carefully engineered prompts to maintain coherence. More than anything, this project was driven by passion—ensuring that the responses remain as true to history as possible.
I’m sharing with you a glimpse of my experience—the closest thing I’ve achieved so far to talking to a historical figure: Siddhartha Gautama.
-
- Facebook Post
-
- Part 1
-
- Part 2
-
- Part 3
-
- Part 4
