The AAP's recommendation against screen time before 18 months is based on a body of evidence showing that babies learn from people, not screens. The key finding is the "transfer deficit" — babies under about 2 years old have difficulty transferring what they see on a screen to real-world understanding. A baby who watches a video of someone hiding a toy can't find the toy themselves, while a baby who watches the same action in person can (Anderson & Pempek, 2005).
This doesn't mean screens are toxic in small doses. The research showing negative effects — language delays, attention problems, sleep disruption — is based on heavy, habitual use (typically 3+ hours per day). Studies on brief, occasional exposure in otherwise stimulating environments don't show the same effects.
The displacement hypothesis explains most of the concern: the problem with screen time isn't primarily what screens do to babies — it's what babies aren't doing while they're watching. Every minute on a screen is a minute not spent in conversation, interactive play, or physical exploration. And those activities are what drive development in the first two years.