I have been in the field of AI for only a few months, but I'm already mesmerized by its capabilities and its potential. What fascinates me most is AGI — one of two subsets of AI, alongside artificial superintelligence (ASI), that have not yet been "invented." That fact alone is what drew me in, and it's why I decided to contribute to building AGI into what it should truly be. Once realized, I don't think AGI will become its own isolated industry. Like electricity or the internet before it, it will act as a general-purpose technology implemented into every sector, gradually transforming each one.
When I started studying AI, I expected to spend most of my time on narrow, task-specific systems — models that translate language, recommend products, or detect fraud. What I didn't expect was how quickly I would become preoccupied with the gap between what these systems can do and what a human mind does without even trying. A child can glance at an unfamiliar animal once and recognize it for life. A chess novice can watch a handful of games and start predicting outcomes with real skill. The systems I was learning to build needed thousands or millions of examples to approximate a fraction of that. That gap is what pulled me from simply studying AI toward wanting to help push it toward something more general. I no longer think of this as a side interest — it's the direction I want my work in this field to take.
How to actually get there is a question I've found far from settled, and wrestling with it has probably taught me more than anything else. My instinct is that AGI will be invented by studying the human brain more closely: its architecture, its efficiency, and the way it learns from so little. That instinct isn't just a hunch — it comes from a specific asymmetry I don't think has been explained away yet. The brain solves problems most AI systems still struggle with using a fraction of the data and a fraction of the power, and no one has shown that this is simply a byproduct of scale rather than of something about the architecture itself. I think this has to be pursued deliberately, because without a guiding model to work from, AGI could remain decades, if not centuries, away. But I've also come to understand that this view isn't the only one taken seriously in the field, and it wouldn't be honest of me to pretend otherwise.
A large part of the AI community believes the opposite: that the biggest breakthroughs of the last decade — the transformer architectures behind today's large language models — succeeded specifically by moving away from biological realism, not toward it. Backpropagation, the algorithm most of these systems use to learn, isn't believed to resemble anything happening between real neurons at all. In that camp's view, progress has mostly come from scale: more data, more compute, more parameters, layered onto fairly simple designs. Still others sit in the middle, working on hybrid neuro-symbolic systems that pair learned pattern recognition with structured, rule-based reasoning, trying to capture the best of both approaches. And there are researchers building hardware and models directly inspired by neuroscience — neuromorphic chips, spiking neural networks, predictive-coding architectures — who are, in effect, trying to prove my instinct right through engineering rather than argument.
The strongest version of the counterargument, I think, is that biological plausibility and performance might simply be orthogonal — that evolution optimized the brain for energy efficiency and survival under constraints that don't apply to silicon at all, so there's no real reason a system built on entirely different principles couldn't outperform it anyway. I take that seriously, and I can't fully rule it out. But that argument only shows brain-inspired approaches aren't necessary, not that they're the wrong bet. Given how much of the remaining gap between narrow AI and general intelligence still looks like a sample-efficiency and energy-efficiency problem, the fact that one system already solves both at once feels like a place worth digging into, not a coincidence worth ignoring. I don't think anyone actually knows yet which path gets there first, and I've stopped expecting a clean answer. But of all of them, the brain-inspired path is the one I keep coming back to, because it's the only one that treats the existence proof already sitting in every human skull as the thing worth reverse-engineering.
What fascinates me most, specifically, is the brain's pattern-recognition ability. It can identify animals at a glance, read a stranger's expression in a fraction of a second, and predict how a situation is likely to unfold almost instantly — all from remarkably little input, and all while running on roughly the power of a dim light bulb. I'm genuinely disappointed that today's AI still falls short of that kind of efficiency, even when it matches or beats human performance on narrow benchmarks. To build a system that outperforms the human brain in general — to actually bring AGI to life — I think we need to study how the brain achieves this with so few examples and so little energy, and translate that understanding into how we design AI, rather than simply scaling our way past the problem.
To make that concrete, I try to picture what it would actually look like once it exists. Imagine an AGI system in medicine that reads a patient's history the way a doctor's eye catches a face in a crowd — noticing a pattern scattered across years of routine visits and flagging a disease before it fully develops, not because it was trained on that exact case, but because it generalized the way a brain does. Imagine a tutor that reads a student's confusion the way a person reads a room, adjusting its explanation in real time instead of running through a fixed script, because it actually understands where the gap in understanding is rather than pattern-matching to a training set. Or imagine a research assistant that can look at a stalled area of science — materials for better batteries, say, or an understudied disease — and propose the kind of intuitive, cross-domain leap that today only comes from a rare human insight, arrived at only after decades of specialized training. None of these are things I expect to see next year. But they're the kind of everyday enhancement I picture when I think about what AGI is actually for — specific enough that I can hold myself to them, rather than just gesturing at "AI will change everything."
Right now, AGI is still just a hypothetical term, defined more by what it would need to do than by anything that currently exists. That gap between imagining it and actually building it is what disappoints me most — but it's also what drives me. I don't think I'll resolve the debate between the scaling camp and the brain-inspired camp on my own, and I'm not sure anyone will resolve it cleanly before AGI actually arrives. But I'd rather spend my time inside that uncertainty, working toward an answer, than admire the question from a distance. I want to help close the gap between the hypothetical and the real, and to see AGI move from a term in a paper to something that quietly strengthens everyday life for Homo sapiens.