The Hacker News post by almustakim47 surfaces a question many developers whisper about: should we fear artificial intelligence, or should we work on it? A colleague once told the author, "You will enjoy exploiting nature and taking the job market." The remark landed because it carried weight—there is truth in it, and with that truth comes responsibility. AI is generating real problems across ecological, economic, and psychological dimensions. Data centers consume power. Job markets feel saturated in some sectors. Relationships with AI-driven platforms contribute to breakups, anxiety, and in tragic cases, suicide. Some experts warn the technology could one day destroy civilization or render humans irrelevant. Yet the same technology has solved significant scientific problems, made learning more accessible, and holds promise for curing cancer and stroke. The article does not minimize the risks, but argues that fear alone solves nothing.
A Philosopher's Train Problem
The author frames the dilemma with a thought experiment. Imagine standing before a train. A machine in your hand can save ten people—your father, your mother, or your lover—but pressing it carries a cost. If you press the machine and cannot break the chain within five minutes, you, roughly eight hundred people in your vicinity, and those ten souls all perish. Researchers estimate a 10-to-20 percent chance you cannot break the chain, and an 80-to-90 percent chance you can. The question becomes pressing: will you press the button to save the ten, or not?
The analogy extends to AI. A non-aligned system could extinguish humanity, researchers suggest, with roughly the same 10-to-20 percent failure probability. But AGI could also solve cancer, where millions die each year. The author's grandfather was among those victims. Skeptics may note the extinction probability lacks a mathematical theorem, but the same skepticism applies to the benefits. The analogy returns to a sobering reality: the button has already been pressed. AI is already deployed, weights are public, and no export control can pull them back.
The Button Already Got Pushed
The unstoppable nature of the technology becomes clear when supply chains and geography are considered. Chinese open models are reported to be about seven months behind the U.S. peak, and the U.S. imposed export controls on two top models in June 2026—controls that were later lifted. Once a model's weights are public, they cannot be recalled. Attribution becomes difficult: how can anyone guarantee they will know if a group is developing systems in secret? The author argues the technology's momentum makes prohibition impossible. Even if many countries halt development, others will continue.
This reality sorts people into three groups. The first consists of those trying to use, control, or grow the technology so harm is minimized and progress continues. The second comprises those who simply do not think about it. The third are those actively trying to stop it, who are against it. The division is not as sharp as it sounds—reality is messier—but the categorization highlights the available postures.
Three Choices, One Path Forward
If the technology cannot be stopped, the author concludes there are only three choices for anyone paying attention. One can do something about it—take advantage of AI, try to control it, steer its growth. One can refuse to engage and simply not care. Or one can be against it, resist, spread fear or hate speech, and accept the technology's dark possibilities. The author rejects the third as irrational. If humans created the problems AI introduces—carbon emissions from data centers, job disruption, psychological harm—then it follows those same humans must solve them. Fear and resignation do not protect the many; only active engagement does.
The numbers reinforce the argument. The chance that humans fail to control AI sits around 10 to 20 percent on average. The chance that AGI could save lives by curing cancer, stroke, or other major diseases is substantial. Even if the extinction risk is real and unproven, the documented toll of cancer and stroke is a concrete mortality figure. The rational position, the author writes, is the first: work on the technology, try to tame it, apply it to the world's hardest problems. That is the only path that protects the largest number of people.
Why Working on AI Beats Fearing It
The article's core thesis is that working on AI is a responsibility, not a resignation. Problems such as ecological strain from compute, economic displacement, and psychological distress from human-AI relationships are real, but they are also engineerable. Data center efficiency improves. Job transition programs emerge. Guardrails for mental health around AI companions can be designed. Each of these outcomes requires people to build, test, and iterate rather than simply fear the unknown.
Moreover, the benefits already appearing—accelerated scientific discovery, more personalized and accessible education, early disease detection—demonstrate that the technology's upside is not theoretical. The author's personal story about his grandfather dying of cancer underscores why the gamble feels worthwhile. If there is a strong chance AGI could prevent millions of deaths, the calculation favors investment and control over abandonment.
The piece ends where it began: with a choice. The button is already pressed. The question now is whether the hands on the machine will steer it toward cure, safety, and progress, or whether they will withdraw and hope the problem resolves itself. The author's answer is clear: engage, control, and build—for the sake of the ten, the 800, and everyone in between.