I have been following the recent warnings from AI experts, including Jacob Coxon and others, about where AI could take humanity in a few years, potentially to the point of annihilation. I understand the fear. When you work at the frontier of AI, you witness capabilities emerging at extraordinary speed. You see systems that can reason, code, plan, use tools, and increasingly act with autonomy. If you are close enough to witness that progression every day, it is understandable that the future can begin to look frightening.
But I keep coming back to a different question. What if being close to AI does not necessarily mean being close to the human world? This is not about intelligence. It is about truly lived experience.
One can spend years studying intelligence, optimization, agents, scaling, and model behavior without necessarily experiencing the full complexity of human existence. One may understand algorithms exceptionally well without understanding the desperation of someone fighting simply to survive. One may understand optimization without experiencing the trauma of losing everything or the uncertainty of rebuilding a life from nothing. One may understand rational decision-making without understanding the irrationality of love, the power of hope, the ability to forgive, or the extraordinary resilience of people who continue living when every rational calculation tells them they should give up. These experiences teach us something that no benchmark can easily measure. Human beings are not optimization problems.
This is one reason parts of the AI existential-risk debate sometimes feel strangely detached from the human world. AI experts have seen the extraordinary power of AI. But proximity to AI does not automatically provide proximity to something equally extraordinary, the complexity, suffering, irrationality, and resilience of the human world. That distinction matters.
I am not suggesting that existential AI risk should be dismissed. The possibility of losing control over increasingly capable systems deserves serious attention. Our understanding of that risk can also become distorted when we examine the future primarily through the lens of AI capability rather than human reality. The real danger may be much more familiar, humans using increasingly powerful AI to amplify human weaknesses such as fear, greed, and the pursuit of power. AI could do the same, only at a scale and speed we have never experienced before.
The AI conversation needs a little less fascination with whether AI will become a god, and a little more humility about whether humans are ready to become the operators of something this powerful. The defining challenge of the AI era is whether human wisdom can keep pace with AI capability. That is the conversation we should be having.