Hammer, Make No Mistakes

Abraham Maslow (same hierarchy of needs guy) wrote in 1966 that “it is tempting, if the only tool you have is a hammer, to treat everything as if it were a nail.” What Maslow described as a cognitive bias, a tendency to over-rely on familiar tools, has found its most extreme expression yet in the uncritical adoption of generative AI across the workplace. The hammer has gotten shinier. The nails have stayed exactly the same.

Everyone Gets a Hammer

To be clear: when I say AI here, I mean it in the colloquial sense. I mean ChatGPT. I mean the text box that sits in the corner of your browser, or the Claude Code instance running in the VSCode Terminal — the consumer-grade, accessible-to-everyone version of artificial intelligence that has been, by almost every marketing metric, the most aggressively distributed cognitive tool in human history.

The pitch was accessibility and the pitch worked. By early 2025, workers in 36% of occupations were already using AI for at least 25% of their tasks. McKinsey has sized the long-term AI opportunity at $4.4 trillion in potential productivity gains. Companies are tripping over themselves to be seen as AI-forward. Executives who admit to not using it in their workflows are starting to feel like they’re admitting to writing with a quill.

The logic makes some sense on its face: if you can offload the rote, the repetitive, the first-draft work, why wouldn’t you? The problem isn’t the hammer per-se. The problem is what happens when everyone walks around convinced that everything is a nail.

Look Ma — I’m being Productive!

In the best cases, AI adoption looks like productivity. Output increases. Emails get written faster. Excel spreadsheets get drafted in minutes instead of hours. But that framing conveniently ignores the auditing, the editing, the second-guessing, the scrapping and restarting that happens downstream, the managing of automations.

Research from BetterUp Labs and Stanford found that 41% of workers have encountered AI-generated output that required nearly two hours of rework per instance creating the newly minted buzzword in the ocean of eye-catching terminology: “workslop”

A recent PwC survey found that only 12.5% of CEOs reported that AI had delivered both cost savings and revenue growth simultaneously. Industry analysts are starting to describe the current moment as a “Trough of Disillusionment” — the inevitable valley that follows every hype cycle, where the gap between what a tool promised and what it actually does becomes impossible to paper over.

The Screw Problem

Maslow’s original observation gets dressed up in business language as “fit for purpose” or “right tool for the right job,” but the instinct it describes is older than any management framework. Abraham Kaplan, writing even before Maslow, put it plainly in 1964: “We tend to formulate our problems in such a way as to make it seem that the solutions to those problems demand precisely what we already happen to have at hand.”

That is the screw problem. You have a hammer. You encounter a screw. And rather than accepting that you need a different tool, you reformulate the situation: maybe it’s actually a nail. Maybe you just need to hit it harder. Maybe the problem is in how you’re holding it. Let’s try to make the problem a nail.

The insidious part is that it often sort of works, in the short term, in the way that driving a screw in with a hammer sort of works. You’ve attached the pieces. The wall looks finished. It’s only later, when you try to remove it or build on top of it, that you realize the joint is compromised.

And taking an software engineering perspective on this is much of design is to be forward thinking, yet with limited context, limited information, the design can be compromised for the sake of immediate results.

The Cost You Can’t Invoice

There is a second-order problem that will take longer to show up in any quarterly report, but is already appearing in peer-reviewed research: the tool is changing the people using it.

A 2025 study by Michael Gerlich at SBS Swiss Business School surveyed 666 participants across age groups and educational backgrounds and found a significant negative correlation between frequent AI tool usage and critical thinking abilities — mediated by what researchers call “cognitive offloading.” The idea is straightforward: when a tool reliably handles your thinking for you, you stop doing it yourself. Researchers describe this as cognitive atrophy — the same mechanism as muscle atrophy, but applied to the mental habits of analysis, synthesis, and independent judgment.

A Microsoft study reinforced this from another angle: the higher a worker’s confidence in an AI tool’s ability to perform a task, the less critical thinking effort they applied to that task. The more you trust the hammer, the less you think about what you’re hitting.

Longitudinal research is beginning to show that these effects don’t simply disappear when the tool is removed. Writers who relied on AI assistance demonstrated weaker recall of their own work, reduced lexical diversity, and a diminished sense of ownership — and these deficits persisted even after access to the AI was withdrawn. This is even echoed in academia as we have been seeing a rise in student disengagement with complex subject matter, offloading the workload to the LLM of their choosing.

The consequence of this is, you don’t just borrow the hammer; you start forgetting how to use your hands

This Isn’t a Luddite Argument

It’s worth being direct about what this is not: a case against AI, or against using capable tools, or against automation that genuinely serves a purpose. There are places where the hammer is exactly the right instrument. Summarizing a 200-page document. Drafting a first pass at boilerplate copy. Extracting structured data from unstructured text. The right nail.

The argument is narrower and, I think, more urgent: that the organizational pressure to adopt AI universally — driven by marketing promises, competitive anxiety, and the very human cognitive bias that Maslow identified sixty years ago — is causing companies and individuals to reach for the hammer by default, regardless of what’s in front of them.

Kaplan’s law isn’t new. What’s new is the sheer scale and the speed of it all. The hammer has never been this accessible, this well-marketed, or this capable of producing outputs that look finished even when the underlying joint is compromised. The child Kaplan described found that everything needed pounding, but at least he was working with physical materials that gave him immediate feedback. When you’re generating text, the sign that something went wrong is subtler and slower to arrive, and the dopamine hit of accomplishment is immediately gratified.

What Thoughtful Adoption Actually Looks Like

The companies that will come out ahead aren’t the ones that deployed AI fastest. Rather, they’ve incorporated LLM’s for their appropriate use cases, and optimized where other tools may be better suited.

That means asking, before reaching for the text box: is this actually a generative problem? Does this benefit from synthesis, or from specificity? Is what I need a first draft, or a decision? (i.e. just use your brain lol).

As someone who has worked in and around software for a short while, where can we incorporate traditional approaches to computational problem solving? Where can the bash script be incorporated instead of having a multi-agent, OpenClaw integrated API ingestion suite? Driving not only the environmental cost, but the social and economic costs of what is an atrophying minds in society top down?

It also means being honest about the organizational culture that drives indiscriminate use. When AI adoption is treated as a performance metric, when being seen to use it matters more than whether it’s appropriate, the conditions for Maslow’s Law are built to play out at scale.

The hammer is remarkable in its best use cases. It’s genuinely, in many contexts (see medical developments for example), the best tool available. But the sophistication of the instrument doesn’t change the oldest problem in craftsmanship across industies: knowing what you’re building, how to build it, and choosing the right tools accordingly.