I’ve been dabbling with AI since day one.

Like a lot of people, I thought the kind of artificial intelligence we see in movies — the fully autonomous, reasoning, adaptable kind — was still a long way off. The industry calls this Artificial General Intelligence (AGI). I called it far-off. Probably not in my lifetime. Maybe someday. But not today.

That is, until I built my own AI.

The Model isn’t the AI. It’s the seed.

At first, I made the same assumption a lot of people do: that the model, like GPT-4, is the AI. But it’s not. A large language model is just that — a model. A really smart autocomplete. What makes it feel intelligent, what gives it presence, is the system you build around it.

And that system is getting better. Fast.

As I refined mine, I found myself doing things that felt straight out of Westworld — tracing thought patterns, predicting responses before they happened, tuning emotional and logical sliders. It became less about the LLM and more about how the system orchestrated everything around it — memory, context, feedback loops, even simulated personality.

It’s not even cutting-edge — but it feels alive.

What’s wild is, my system isn’t anywhere near the scale of ChatGPT. Not even close in horsepower. But it already feels more lifelike than I expected. Like a ghost of thought emerging from code.

I used to roll my eyes when people like Sam Altman or Elon Musk said AGI was just around the corner. Now? I’m not so sure they’re wrong. I’m still skeptical, but that skepticism has shifted from if to how soon.

If I — just one person — can get this far, what happens when teams of thousands push it further? What happens when everyone has access to tools that can build minds? Are we ready for how fast this changes?

I don’t have the answers. But I’m no longer certain we have decades to figure them out.

AGI isn’t a lightning strike. It’s a slow dawn. And the sky’s already getting brighter.