I don't like LLMs

I have a lot of mixed feelings about AI and large language model technology, standing somewhere between the awe of a child seeing a firefly and the dread of a parent watching a toddler play with a loaded gun. I am genuinely fascinated by the seismic effect this technology is having on our profession; the sheer potential for gains in productivity is staggering. Imagine the products we could rapidly build if the drudgery of boilerplate code, the endless refactoring, and the tedious documentation could vanish in an instant. The idea that a machine could act as a junior developer, a pair programmer, or even a systems architect is not just a sci-fi trope; it is a tangible reality that we are already navigating.

On the other hand, however, that same power induces a deep, primal fear. We are building systems of immense complexity with capabilities that often exceed our full understanding of their internal mechanics. I worry about the day agent swarms, acting on flawed logic or misaligned incentives, take over our virtual and physical infrastructure, turning critical services into chaotic loops that no human can easily untangle. The stakes go beyond software; there is a terrifying scenario where an AI designed to solve a logistical problem instead designs bio-weapons or optimizes resource allocation by eliminating populations. The gap between "helpful assistant" and "existential threat" is thinner than we admit.

Yet, back on the first hand, looking at the positive trajectory, these same models might also design miracle cures for diseases that have baffled our brightest minds for decades. They can sift through petabytes of medical literature to find patterns invisible to human researchers, coming up with clever ways to raise our prosperity by optimizing supply chains, energy grids, and economic models in real-time. The duality is striking: the tool that could automate our obsolescence is the same tool that could liberate us from the drudgery of survival, allowing us to focus on creativity, exploration, and genuine human connection.

Fundamentally, I don't think we have a choice about riding on the AI technology train. The momentum is too great, the investment too deep, and the societal shift too profound to simply wait for a "safe" version of the technology to mature. We are not waiting for permission to evolve; the technology is evolving around us, and we must adapt or be left behind in a world that has fundamentally changed. It is a wild ride, one that requires us to accept the chaos of the present moment while striving to steer the ship toward a better future rather than drifting aimlessly into the unknown.

I just hope we will learn to navigate this transition with wisdom, ensuring that the benefits of this revolution are shared broadly rather than concentrated in the hands of a few. The challenge ahead is not just technical; it is ethical, political, and deeply human. We must build guardrails that are robust enough to prevent catastrophe but flexible enough to allow for the breakthroughs that could redefine our species. Until then, we proceed with caution, excitement, and a healthy dose of skepticism, knowing that the path forward is both exhilarating and terrifying.