Designing for Reflection in the Age of AI
Don’t Make Me Think by Steve Krug is a foundational book on web usability that emphasizes designing websites and apps so intuitive that users barely have to think to use them. Krug argues that good design should be self-evident, relying on clear visual hierarchy, familiar conventions, and minimal distractions to help users achieve their goals quickly. He stresses that users skim rather than read, make quick decisions, and often muddle through rather than follow instructions, so interfaces should be simple, forgiving, and focused on usability testing rather than perfection.
Steve Krugās Don’t Make Me Think was foundational to the user-centered design principles that shaped Web 2.0, emphasizing simplicity, clarity, and minimal cognitive effort. These ideas influenced the rise of social media platforms that prioritized ease of use, instant feedback, and addictive interfaces. As a result, users could engage effortlessly, often mindlessly. While this made the web more accessible, it also ushered in an era where critical engagement was displaced by frictionless scrolling and superficial interactions. Arguably, the mantra of “donāt make me think” became a double-edged sword: it enhanced usability while encouraging passive consumption over thoughtful participation.
As the unintended consequences of seamless digital experiences become more apparentāaddiction, misinformation, and disconnectionādesigners are increasingly recognizing the value of friction in user experience. These intentional pauses or interruptions in automation re-engage the userās attention and bring their reflective mind back into the loop. Rather than optimizing every interaction for speed and ease, friction-based design introduces moments for choice, context, or reconsideration. Examples include double-checking before posting, taking a mindful pause before continuing a scroll, or offering deeper context behind a notification. By making users thinkānot in the obstructive way Krug warned against, but in a conscious and intentional wayāfriction becomes a tool for ethical, human-centered design in an age that too often rewards mindless engagement.
Good friction in payment experiences introduces intentional pauses that enhance user safety, trust, and decision-making. For example, confirmation prompts before finalizing a purchase help prevent accidental or impulsive spending, while address and card verification steps add a layer of security that reassures users. Multi-factor authentication, especially for large or unusual transactions, introduces a brief delay that significantly reduces fraud risk. Review screens that summarize items, costs, and terms give users a final chance to catch errors or reconsider. Even budget alerts or spending warnings can nudge users toward more mindful financial behavior. These design choices slow the process just enough to bring the userās conscious mind back into the loop, turning friction into a feature, not a flaw.
Some critics claim that AI will do our thinking for us and ultimately make us dumber, pointing out that AI systems often hallucinate, confidently producing false or fabricated information. For example, an AI might generate a plausible-sounding academic citation that doesn’t actually exist. Minimizing such errors is crucial, but it’s also worth noting that error is not a disqualifier of intelligenceāitās part of it. In Knowledge and the Flow of Information, philosopher Fred Dretske argues that the capacity to misrepresent is essential to genuine representation. A mental or informational state can only count as representing something if it can get it wrong. We accept that humans err and build systems (legal, scientific, educational) that account for this. So why not extend the same adaptive approach to machines?
The issue arises from holding AI to an outdated ācommandā model of automation, where computers are expected to execute perfectly defined tasks with precision. But AI belongs to a ācollaborateā model: it processes and proposes, while the human remains in the loop, interpreting, validating, and deciding. AI can do much of the heavy lifting of information processing, but ultimate accountability still rests with people. In this light, the challenge of designing with AI isnāt to eliminate thinking, but to prompt it at the right moments. A fitting design ethos for our time might flip Krugās classic title on its head: Make Me Think.
