Why 'AI Is Just a Tool' Falls Apart

Calling AI “just a tool” misses the point, according to a widely discussed post on Frank Computer that the author updated in 2026 after months of traffic and debate. The post takes aim at the familiar defense that “AI is just a tool — it matters how you use it,” arguing that this framing is too narrow to describe technologies that reshape culture, labor, infrastructure, and even human behavior.

The author’s core claim is simple: tools are never neutral. A car is not only about how a person drives it; it also changes cities, climate, and safety norms. By the same logic, the effects of modern generative AI cannot be reduced to individual use. The post argues that ethics also depend on how AI is built, how it is distributed, what resources it consumes, and what kinds of systems it reinforces.

The essay draws on Heidegger’s idea of “en-framing”, or the notion that technologies shape who people become through their design. A chair, the post argues, does not merely allow sitting; it instructs the body to “sit still, face forward, and behave.” AI, in this view, exerts a stronger pressure still: toward convenience, dependence, and the avoidance of thought, imagination, and difficulty.

Struggle, drudgery, and what automation removes

A major theme is the difference between meaningful struggle and pointless barriers. The author argues that some friction should be eliminated — like a curb that blocks accessibility — while other forms of effort are central to human life, like making art, writing, or working through a hard problem.

×

The post uses that distinction to criticize the way AI is marketed as a cure for all effort. Offloading writing, art, and thinking to models may remove drudgery, the author says, but it can also erase the very struggle that gives those activities meaning. In that sense, the essay compares AI less to a neutral productivity aid than to an opiate: a technology that numbs experience while hiding its costs.

Recommended reading

Hassabis says STEM makes you 10x better at AI

Environmental damage, data theft, and policy

The post ultimately argues that today’s transformer- and diffusion-based models are “largely bad to use, especially now, and in most all contexts.” It cites three categories of harm:

• Environmental: pressure on fresh water and energy resources, and accelerated climate change
• Economic: training built on mass scraping, described as “the largest heist in human history”
• Existential: pressure to hand over more of life to optimization and automated systems

×

To support the last point, the author quotes Tina He, who writes that the anxiety produced by these systems may reveal “something vital, unquantifiable, and irreducibly human” that still resists automation. The post does not call for abandoning tools outright. Instead, it argues for policy, economic justice, and stronger guardrails before these systems can be meaningfully aligned with human values.