Think Slow, Act Fast: The Smarter Way To Use AI
Marketing Against the Graingo watch the original →
the gist
David Epstein argues that AI is currently being used to create 'work slop' because teams prioritize tools over problem definition; he advocates for 'thinking slow' by applying constraints—like writing a press release before building—to ensure output actually solves user needs.
The 'Think Slow, Act Fast' Framework
Most organizations suffer from a 'think fast, act slow' pattern, where they rush into execution without defining the core problem. This leads to costly pivots and 'work slop.' David Epstein highlights research from Bent Flyvbjerg showing that successful projects prioritize 'thinking slow'—setting clear strategic boundaries and constraints early. By defining the goal and the target audience before touching a tool, teams can move significantly faster during the execution phase because the 'sandbox' is clearly defined.
Constraints as a Catalyst for Clarity
Constraints are not obstacles; they are design tools. Epstein points to Tony Fadell’s practice of writing a press release at the start of a project to force clarity on the value proposition. Similarly, the 'box' exercise—where teams must fit their product's core value onto a physical package—prevents feature creep. In an era of infinite AI-generated output, simplicity is a competitive advantage. Teams should embrace 'desirable difficulty' by forcing themselves to articulate the problem before seeking a solution, ensuring they aren't just building 'a broom with a laser pointer.'
The Explore-Exploit Cycle
Epstein frames his work as a study of the 'explore-exploit' trade-off. Range focuses on the 'explore' phase—broadening horizons and building a diverse toolset. Inside the Box focuses on the 'exploit' phase—applying those tools within strict boundaries to achieve high-impact results. The failure of many organizations, such as Kodak, stems from an inability to cycle back to exploration, or conversely, an inability to focus on exploitation once a direction is chosen.
AI as a Tool, Not a Brain
AI is currently being misused as a replacement for thinking. Epstein argues that AI should be treated as a 'GPS for your brain'—a tool to accelerate research or synthesis—rather than a generator of first-draft content. Writing is a process of discovery that exposes gaps in one's own understanding; outsourcing this to AI prevents the user from actually learning the subject matter. The 'subtraction audit'—cutting one thing every quarter—is a vital practice to prevent the accumulation of useless features and processes.