Much of my research career has been shaped as much by chance as by design. A stray questionnaire item about childhood trauma led to the discovery that keeping secrets can affect people's health, and ultimately to the expressive writing paradigm. Trying to understand why writing about emotional experiences could be beneficial led to the development of LIWC, a text analysis method that revealed how everyday language can tell us things about people that often defy our intuitions. Small words like "I" and "we" can reveal surprising insights about our health, relationships, and social worlds.
In this talk, I'll trace how a series of unexpected findings led from small laboratory studies to massive real-world datasets and, more recently, to AI methods that are transforming how we study human behavior. Along the way, these developments have raised increasingly uncomfortable questions about the assumptions underlying psychological science. If machines can predict human behavior without necessarily understanding it, what is the role of psychological theory? And how should we be training the next generation of scientists?