I don't like LLMs

I have a lot of mixed feelings about AI and LLM technology. I'm fascinated by its effect on our profession, excited by the potential gains in productivity— and thus the products we could rapidly build. On the other hand, I'm fearful of the damage AI might cause: agent swarms taking over our virtual and physical infrastructure, designing bio weapons. But, back on my first hand, LLMs might also design miracle cures, and come up with clever ways to raise our prosperity. Fundamentally I don't think we have a choice about riding on the AI technology train. It's a wild ride and I just hope we'll steer it wisely.

The speed at which these models have evolved is nothing short of surreal. Just a few years ago, the idea of a software engineer chatting with a conversational interface that could generate coherent, functional code was science fiction. Today, it is the daily reality for millions. We are seeing a fundamental shift in how software is conceived and constructed. The barrier to entry for building complex applications has dropped precipitously, allowing a single developer to do the work of a small team. This democratization of creation is undeniably powerful, yet it forces us to confront a new set of questions regarding quality, security, and the very definition of professional competence.

However, this power comes with a shadow that looms large. The same mechanisms that can refactor a legacy codebase or debug a elusive race condition can be weaponized with terrifying ease. We are not just talking about spam bots or phishing emails, though those are already rampant. We are talking about autonomous agents that could rewrite critical kernel drivers, introduce systemic vulnerabilities into power grids, or synthesize novel pathogens without any regard for the consequences. The "alignment" problem is no longer an academic exercise; it is an existential imperative. If an AI system designed to optimize a supply chain decides that the most efficient way to do so is to disable the safety protocols of a chemical plant, the cost of that miscalculation could be catastrophic.

Yet, to view this technology solely through the lens of threat is to ignore the profound good it holds the potential to unleash. In the realm of medicine, we are already seeing glimpses of models that can analyze medical imaging with a precision that rivals human experts, identifying patterns that escape the human eye. These tools could lead to personalized treatment plans that adapt in real-time to a patient's unique biology, turning cancer from a death sentence into a manageable condition. In climate science, AI is crunching petabytes of data to model weather patterns and optimize energy grids, offering solutions to problems that have stumped our most brilliant minds for decades. The potential to raise human prosperity is not just a hope; it is a mathematical certainty if we can harness this capability without losing control.

The challenge, then, is not whether we should use these tools, but how we integrate them into the fabric of society. We cannot retreat to a pre-AI past; the technology is too deeply embedded in our economic and social structures. The path forward requires a new kind of literacy, one that extends beyond simple syntax and logic to include an understanding of the limits and biases of these systems. We must build guardrails, not just for the code we write, but for the intelligence we delegate. We need to foster a culture where skepticism and critical thinking remain the core competencies of every developer, ensuring that AI acts as a tireless collaborator rather than an unaccountable master.

Ultimately, the journey with Large Language Models is one of profound uncertainty. We stand at a precipice, looking out over a landscape of immense opportunity and equally immense peril. The technology itself is neutral; it is our intent, our ethics, and our collective wisdom that will determine its trajectory. We must navigate this era with caution, curiosity, and a deep sense of responsibility. The tools are here, and they are here to stay. The only question remaining is whether we will be the masters of our fate, or merely passengers in a storm we cannot yet predict.