The AI Effect: Why We Always Move the Goalposts of Intelligence

Every time artificial intelligence achieves a milestone once thought uniquely human—defeating a world champion at Go, diagnosing rare medical conditions from scans, or writing a fluent sonnet in seconds—a strange cultural phenomenon occurs.

Instead of marveling at the triumph, our collective reaction is often a muted shrug: “Well, that’s just glorified pattern matching. It doesn’t actually understand anything.”

This phenomenon has a name: The AI Effect.

Coined by computer scientist John McCarthy, the AI effect describes the tendency to dismiss artificial intelligence achievements as “not real intelligence” the exact moment they become commonplace. In short: AI is only considered intelligence until it works. Once a machine masters a task, we stop calling it AI and simply call it software.

Moving the Goalposts of the Mind

For decades, the definition of artificial intelligence has been a moving target. In the mid-20th century, automated arithmetic was considered a marvel of automated thought. Today, pocket calculators do it instantly, and we view them as primitive tools.

Consider the trajectory of chess:

• 1950s: Computer scientists declared that if a machine could beat a human chess grandmaster, it would prove true machine intelligence had arrived.
• 1997: IBM’s Deep Blue defeated world champion Garry Kasparov.
• The Aftermath: Rather than conceding that machines could think, critics argued that Deep Blue merely brute-forced millions of calculations per second using raw computing power, lacking true strategic intuition.

This pattern repeats itself in every domain. Optical character recognition (reading handwritten mail), recommendation algorithms (predicting what movie you want to watch), and speech-to-text transcription were all once cutting-edge frontiers of AI research. Today, they are background infrastructure—invisible, mundane, and thoroughly unmagical.

Why Do We Fall for the AI Effect?

Human psychology plays a massive role in how we perceive technology. We tend to define intelligence through an anthropocentric lens—measuring it by our own subjective experiences of consciousness, emotion, and intuition.

1. The Demystification Trap: When a complex cognitive process is broken down into code, logic gates, and statistical probabilities, it loses its aura of mystery. We mistake knowing how it works for it not being intelligent. By that logic, human cognition—which is ultimately a network of firing neurons and biochemical reactions—might also be dismissed if we could fully map it.
2. The Illusion of Effortlessness: Modern generative models can paint, code, and converse in milliseconds. Because humans require intense concentration and years of practice to write a symphony or draft a legal brief, we project our internal struggle onto the task. When a machine produces the same output instantly and without sweat, we assume it must be cheating rather than thinking.

The Danger of Dismissing the Machine

The AI effect isn’t just a quirky psychological quirk; it has real-world consequences.

When we write off AI achievements as “mere statistics” or “stochastic parrots,” we risk falling into two opposite traps: underestimating its societal impact or anthropomorphizing its capabilities.

• Underestimation: Dismissing AI systems as dumb tools blinds us to their very real capacity to disrupt labor markets, spread misinformation, and shift power dynamics. Even if an AI doesn’t “feel” or “understand” in a human sense, its outputs can still alter human lives profoundly.
• Overestimation: Conversely, because we expect machines to think like humans, we can easily be fooled into trusting them blindly when they hallucinate facts or exhibit biases, forgetting that they are fundamentally pattern-recognition engines operating without moral agency.

Redefining Intelligence in a Post-AI World

Ultimately, the AI effect forces us to confront an uncomfortable philosophical question: What is intelligence anyway?

For centuries, we defined humanity by our ability to reason, calculate, create art, and use language. Now that silicon chips can replicate—and sometimes exceed—those abilities, we are forced to look deeper.