AI Should Not Be Called 'Intelligence' Because It Lacks Essential Human Qualities
Source: Charles Yu. "Don’t Call It 'Intelligence' - The Atlantic." March 5, 2026. www.theatlantic.com
The Gist
The author argues we shouldn't call AI 'intelligent' because real intelligence involves much more than just answering questions expertly. True intelligence includes emotions, curiosity, learning from failure, and understanding things we can't put into words - none of which AI has.
Conclusion
What AI can do should not be called 'intelligence' because it fundamentally differs from human intelligence in ways that make the term a misnomer
Premises
- Current AI definitions of intelligence (like Altman's 'Ph.D.-level expert in anything') only capture a narrow subset of what intelligence actually encompasses
- True intelligence includes emotional intelligence, tacit knowledge, embodiment, curiosity, and the ability to experience wonder, grief, and awe - none of which AI possesses
- AI is fundamentally an 'answer machine' while humans are 'question machines' - and questions are essential to intelligence
- Human intelligence emerges through failure, struggle, and developing one's own voice, while AI bypasses this essential process
- Most human knowledge is tacit knowledge that cannot be expressed symbolically or mathematically, yet AI only processes explicit, tokenizable knowledge
- Calling AI capabilities 'intelligence' causes us to underestimate human abilities and conflate technological capability with human attributes
Assumptions
- Intelligence necessarily includes emotional and experiential components, not just cognitive processing
- The struggle and failure involved in developing human voice/intelligence is essential, not merely incidental
- Tacit knowledge represents a fundamentally different and larger category than explicit knowledge
- Questions and curiosity are more fundamental to intelligence than providing answers
- Human intelligence cannot be meaningfully separated from embodiment and emotion