Sotomayor: AI that forecasts Supreme Court rulings is “a bad thing” — “we’re way too predictable”

It has been reported that Justice Sonia Sotomayor criticized efforts to use artificial intelligence to predict Supreme Court decisions, calling such forecasting “a bad thing” and saying the projects show “we’re way too predictable.” The comment, surfaced in a Reddit thread, allegedly came during remarks in which the justice warned about the implications of turning judicial behavior into a data problem. Take it with a grain of salt — the sourcing is social-media-based — but the line landed hard.

What she said

According to the post, Sotomayor argued that reducing judicial reasoning to patterns ripe for machine prediction risks eroding the institution’s mystique — and maybe its independence. “We’re way too predictable,” she reportedly quipped, which is a provocative admission coming from a sitting justice. If true, it raises a thorny question: should courts be opaque to preserve deliberative space, or transparent because accountability demands it?

Why it matters

Predictive models for court outcomes aren't just academic toys; startups and researchers have built systems that claim respectable accuracy on past rulings. So the worry is real. If AI can reliably forecast votes, what does that do to case selection, lawyer strategy, or public trust? It could chill certain litigants or, worse, incentivize strategic litigation to game predictable judges.

Reaction and context

Expect quick pushback and serious debate. Legal scholars and technologists have long argued about the benefits of data-driven transparency versus the dangers of mechanizing judgment. Whether Sotomayor actually uttered those words exactly as quoted, the Reddit-sourced line — “a bad thing” — has lit a fuse. Can a court be both principled and unpredictable? That’s the heart of the argument, and it’s one the AI era just made a lot harder to dodge.

Sources: reddit