Resisting Cognitive Atrophy While Scaling with AI

The efficiency of Generative AI seems almost like an irresistible siren song. However, recent research highlights the hidden professional costs that can accompany this efficiency. In "AI Tools in Society: Impacts on Cognitive Offloading and the Future of Critical Thinking", Michael Gerlich demonstrates how delegating mental tasks to AI can lead to "cognitive offloading," a process that may dull critical thinking abilities over time (Gerlich p. 2). Complementing this, André Barcaui’s "ChatGPT as a Cognitive Crutch: Evidence from a randomized controlled trial on knowledge retention," measures the practical impact of this phenomenon on knowledge retention. By comparing AI-assisted learning with traditional methods, Barcaui confirms that the mental shortcuts Gerlich warns about lead to a significant deficit in actual learning and directly impairs a practitioner’s ability to retain information long term (Barcaui p. 4). Together, these works suggest that an uncritical reliance on AI is more than a change in workflow; it is a potential risk to long-term professional edge.

One of the more notable findings is the impact of AI on memory. Barcaui’s study found that students who used ChatGPT as a study aid scored lower (57.5%) on a surprise retention test 45 days later than those who used traditional methods (68.5%) (p. 18). This 11% gap suggests a difference in how our brains encode information when using AI.

The data reveals a "significantly steeper forgetting curve" for AI users (Barcaui p. 15). This is likely because AI eliminates "desirable difficulties", or the productive struggle of manual research and drafting that anchors knowledge (p. 4). As Barcaui notes, AI assistance can affect the quality of learning at the outset, not just the ability to retrieve information later (p. 14). For a lawyer, using AI to quickly "get up to speed" on a complex regulation might result in a more superficial understanding that is harder to recall during a high-stakes, unscripted negotiation.

Part of the value of an in-house lawyer lies in one's ability to analyze, evaluate, and synthesize information to make reasoned decisions. However, Gerlich’s research identified a strong negative correlation (r = –0.68) between frequent AI tool usage and critical thinking benchmarks (p. 13 & 14).

This trend is driven by "cognitive offloading", which is the habit of delegating analytical tasks to an external tool and can reduce deep, reflective thinking (Gerlich p. 24). Gerlich warns that this can create a feedback loop where increased reliance on AI exacerbates a decline in independent analysis (p. 14). When we offload the initial analysis of a legal issue to an AI, we may save time, but we also bypass the mental exercise required to maintain a sharp problem-solving edge. Over time, this could lead to a workforce that is highly efficient but less practiced in independent evaluation (p. 3).

A subtle but insidious risk is what the researchers call "borrowed competence" (Barcaui p. 17). This is a metacognitive blind spot where a professional might mistake an AI’s fluent output for their own mastery of a subject.

Because AI provides immediate, polished answers, it can create a false sense of security. Gerlich’s study notes that as users develop greater trust in these tools, they may be less likely to scrutinize recommendations thoroughly (p. 22). In a legal context, over-reliance on an AI's summary can lead to "cognitive laziness," where a practitioner might miss subtle biases or logical inconsistencies in the algorithm's recommendations.

Efficiency is a compelling goal, however, an uncritical pursuit of it can mask the dulling of our analytical edge. When we routinely delegate complex analysis to an algorithm without independent thought or review, we risk trading long-term sharpness for immediate convenience. Expertise is not a permanent state but a level of cognitive fitness we must actively maintain (see Competence 3.1-2 for the required standard of a competent lawyer).

If we consistently bypass the heavy lifting of critical inquiry and deep research, our problem-solving reflexes can weaken. This becomes a real vulnerability during high-stakes situations where nuanced and unscripted judgment is required. The challenge for modern in-house counsel is to harness the speed of AI without passively surrendering to its ease. We must not let our core skills atrophy in the quest for a faster turnaround. Sustained success relies on treating artificial intelligence as a capable assistant while keeping our own analytical abilities fully engaged.

ChatGPT as a Cognitive Crutch: Evidence from a Randomized Controlled Trial on Knowledge Retention  André Barcaui November 2025 AI Tools in Society: Impacts on Cognitive Offloading and the Future of Critical Thinking  January 3, 2025 Michael Gerlich