Why aren't smart people happier? (2022)

Intelligence-test scores show little relationship with happiness, according to meta-analyses and a large nationally representative UK study cited in the source. An analysis of 50 years of U.S. General Social Survey data covering 30,346 people found a small negative correlation between vocabulary-test performance and self-reported happiness (r = −.06, p < .001). Lower-scoring participants in the UK study were somewhat less happy, but higher scores did not correspond to substantially greater happiness. The source argues that conventional intelligence tests chiefly measure skill at solving well-defined problems: tasks with stable rules, clear boundaries, finite information, repeatable processes, and indisputable answers. Charles Spearman’s finding that cognitive-test scores tend to correlate positively is widely replicated, but the source contends that it does not establish a single ability applicable to life’s poorly defined problems, such as relationships, careers, parenting, or finding a satisfying life. It further argues that AI progress has focused on well-defined tasks because training data and evaluation require problems to be defined. It characterizes systems such as GPT-3 and DALL-E 2 as performing bounded tasks—next-word prediction and text-image matching—rather than independently resolving poorly defined human questions. The source says wisdom, self-knowledge, creativity, and agency may matter more for such questions, but no established test captures that capability.