The author argues that despite the utility of "AI" tools like Google's AI Overview, society must resist their adoption until significant harms are addressed. The piece focuses on three critical areas of damage.

Environmental harm is staggering: training current models costs up to $100 million, producing massive CO2 emissions, consuming over a billion Watt-hours of electricity, and using a million liters of water. Ongoing data center power consumption equals 100,000 households, with daily water usage matching 50,000 households.

Intellectual harm occurs when developers rely on AI for answers rather than actively engaging with code. The author warns that reviewing AI-generated code leads to superficial checks—hunting for typos rather than deep understanding—eroding the mental model of the systems they are responsible for building.

Human harm is already manifesting as corporations replace workers with AI. The author notes AI creators themselves are shedding programmers, likely to offset massive training costs. Without a post-scarcity society, job elimination threatens livelihoods. The call to action targets executives, managers, consultants, and developers to resist AI integration until harms are eliminated, mitigated, or compensated for.