Why Both Sides of Any Debate Often Make the Same Cognitive Mistake

A recent analysis explores why public debates, particularly around agentic AI, tend to collapse into extreme positions rather than nuanced ones. Forming a well-reasoned opinion demands significant cognitive effort, including understanding a problem, finding precise language, and articulating it clearly — shortcuts are tempting as a result. Broad, imprecise category labels like 'AI' worsen polarisation by allowing opposing parties to believe they are debating the same thing when they are actually talking past each other. Once an opinion becomes tied to personal identity, it hardens under resistance, creating a self-reinforcing loop that resists correction. Social media algorithms further entrench this dynamic by rewarding vivid, polarising content over accuracy and balance, giving creators and platforms little incentive to change.