Is AI making us dumber? | Marvin Liyanage @marvinliyanage

Key Concepts

• Fluency Heuristic: The tendency to judge the accuracy of information based on how easily it is understood.
• Use it or Lose it: The principle that cognitive abilities are maintained and improved through regular use; lack of use leads to decline.
• Outsourcing Thinking: Relying on AI to perform cognitive tasks that would normally be done by a human, potentially hindering critical thinking skills.
• Brain Activity & Cognitive Load: The relationship between brain activity levels and the amount of mental effort being exerted.

The Claim: AI and Cognitive Decline

The central argument presented addresses the widespread concern that Artificial Intelligence (AI), specifically tools like ChatGPT, is negatively impacting cognitive abilities – essentially “rotting your brain.” The speaker directly challenges this claim, drawing parallels to past anxieties surrounding new technologies, citing a 2008 Atlantic article titled "Is Google making us stupid?" as a historical precedent.

Brain Activity: Less Work, Not Less Intelligence

A study from MIT is referenced, demonstrating that brain activity decreases when using ChatGPT. However, the speaker argues this reduction in activity doesn’t equate to intellectual decline. Instead, it signifies a shift in cognitive load – the brain is simply performing less work because the AI is handling certain tasks. This is framed using the analogy of physical exercise: just as muscles atrophy without use, cognitive abilities weaken without regular engagement. The example of using a calculator is provided; delegating long division to a calculator doesn’t make one less intelligent, but frees up mental resources for other tasks.

Critical Thinking & The Fluency Heuristic

The discussion pivots to the specific concern that AI hinders critical thinking. While acknowledging ChatGPT can be used to access and synthesize information (e.g., 20 research papers), the speaker points out that most users don’t utilize it in this way. A key concept introduced is the fluency heuristic, which explains why people tend to equate clear and easily digestible writing with accuracy. This is particularly relevant to AI-generated content, as its polished and coherent presentation can mask underlying inaccuracies or biases. The speaker emphasizes that AI makes mistakes, and users should not assume its output is inherently truthful.

How Students are Using AI: A Breakdown of Usage

Data on student usage of AI is presented. The most common application is as a tutor, which the speaker suggests doesn’t necessarily indicate a lack of independent thought. However, the second most common use – writing entire essays – is identified as a potentially problematic practice, representing a form of outsourcing thinking.

Agreement & The Core Issue

The speaker establishes a point of agreement: AI isn’t inherently making people stupider. The crucial factor is how it’s used. If AI is employed to completely bypass cognitive effort, it can indeed be detrimental. The video concludes with a direct, rhetorical challenge – “You literally just asked ChatGPT” – highlighting the irony of questioning AI’s impact while simultaneously relying on it for answers.

Synthesis

The core takeaway is that AI is a tool, and its impact on cognitive abilities depends entirely on the user. While concerns about reduced brain activity and the potential for outsourcing thinking are valid, they are not inevitable consequences. The fluency heuristic poses a significant risk, as users may be misled by the apparent clarity of AI-generated content. Responsible AI usage involves critical evaluation, independent thought, and a recognition that AI is a supplement to, not a replacement for, human intelligence.