Think Little

In the AI Story

Berry wrote "Think Little" in 1972 as a direct response to the environmental movement's emerging strategy of thinking big—national legislation, comprehensive plans, technological solutions scaled to address global problems. Berry's essay argued this strategy was self-defeating: problems are always local and specific in their manifestation, and solutions imposed from above without understanding ground conditions produce consequences the planners did not anticipate. The essay became a founding text of the bioregional movement and the agrarian revival, influencing figures from Wes Jackson (founder of the Land Institute) to Fred Kirschenmann (distinguished fellow at the Leopold Center). Berry's argument was not anti-government or anti-planning per se—it was pro-specificity, insisting that effective action requires intimate knowledge of the context, and intimate knowledge requires presence at the scale where details are visible.

Applied to artificial intelligence, "think little" becomes a discipline of resistance against scale economics. AI's marginal cost structure (near-zero cost of serving additional users) creates overwhelming economic pressure toward universal tools serving universal needs. Berry would ask: universal according to whom? Designed for what context? With what assumptions about what constitutes valuable work? The tools Segal describes—Claude Code, frontier language models—were designed in San Francisco, trained on predominantly English data, optimized for knowledge-economy workflows. They are sophisticated tools. They are not context-neutral tools. Every design choice—what counts as "good" code, what communication style the model adopts, what problems are considered worth solving—embeds specific cultural and economic assumptions. These assumptions serve some contexts well and others poorly, but the economics of scale make differentiation expensive and cultural homogeneity profitable.

The "developer in Lagos" whom Segal invokes as the moral case for AI democratization illustrates the tension. The developer deserves tools—Berry would not dispute this. But does the developer need the same tools that serve Silicon Valley developers, or does the developer need tools designed for Lagos's infrastructure constraints, the local market's specific needs, the community's particular problems? A universal tool may serve some of these needs. It cannot serve all of them, because serving all would require the local knowledge that no universal system possesses. "Think little" means building tools that serve specific communities rather than all communities, even though the economics push relentlessly toward the opposite. It means accepting that the optimal solution for Lagos and San Francisco are different solutions, and that the difference is not inefficiency but precision.

Berry's prescription for organizations deploying AI: before you scale, understand. Before you generalize, specify. Before you build for millions, build for the person in front of you—the actual student in the actual classroom, the actual worker with actual constraints in an actual place—and make sure what you build serves that person. Not the abstracted "user" of the product manager's imagination but the whole human being in a specific context with specific needs. This discipline is harder than thinking big. Thinking big is intoxicating because it liberates the thinker from messy, resistant, specific details. Thinking little requires staying in the mess, attending to specifics, resisting the abstraction that makes work feel more important than it is. The parent at the kitchen table does not need to think big about AI's economic impact. The parent needs to think little: What does my child need, tonight, at this table? The specific answer to that specific question is more important than any abstraction about the future of work.

Origin

"Think Little" was originally published in A Continuous Harmony: Essays Cultural and Agricultural (1972), written during the period when Berry was most actively engaged with the environmental movement and most critical of its strategic direction. The essay's title inverted the dominant logic of the era—"think globally, act locally" was emerging as the environmental slogan, and Berry was arguing that the formula had it backward. Thinking globally produces abstractions. Acting locally produces results. The sequence should be reversed: think locally, understand specifically, and let the global implications emerge from the accumulation of local successes rather than imposing global solutions that ignore local conditions.

The concept has become more urgent as technology's scale economics have made "thinking big" not merely an option but an economic necessity. Platforms, by definition, serve scale. Code, once written, replicates at zero marginal cost. AI models, once trained, serve millions. The economic pressure is always toward more users, more generality, more reach. Berry's "think little" is not a rejection of technology but a discipline of resistance—the practice of asking, at every decision point, whether the scale serves the work or whether the work is being distorted to serve the scale.

Key Ideas

Right scale enables care. The scale at which a person can observe consequences, adjust to learning, and maintain relationships between intervention and system is the scale at which care is possible—typically smaller than industrial economics will tolerate.

Abstraction replaces attention. Thinking big produces models, strategies, comprehensive plans; thinking little produces presence, observation, adjustment—the latter is slower and less dramatic, but it is what actually works in complex living systems.

Universal tools serve universal abstractions. Every tool designed to serve everyone serves the averaged, generalized user and serves actual specific users in proportion to how closely they resemble the average—making universal tools systematically less useful to populations furthest from the design assumptions.

Local knowledge is non-transferable. The understanding that emerges from sustained presence in a specific place—what this field needs, what this student struggles with, what this codebase will do under stress—cannot be replaced by general principles or AI-generated summaries of best practices.

Small work compounds. Berry's faith—tested over sixty years—is that good work at small scale accumulates into large effects through imitation and adaptation, while comprehensive plans imposed from above produce resistance and unintended consequences that planners did not anticipate because they were not present at the ground.