⚠️ SPOILER WARNING: If you haven’t watched Mercy yet, turn back now.
Okay, so you just finished Mercy and you’re sitting there like “wait… what just happened?” Don’t worry, you’re not alone. That ending is a lot to unpack.
The Setup
Let’s recap: Detective Alex has 90 minutes to prove he didn’t kill his wife. The AI judge—called THEMIS—analyzes every piece of evidence with zero emotion. Throughout the trial, Alex keeps insisting he’s innocent, but the evidence keeps stacking against him.
The Twist
Here’s where it gets wild. In the final 10 minutes, Alex stops trying to prove his innocence. Instead, he questions THEMIS itself.
He points out that THEMIS was programmed by humans—flawed, biased humans. He argues that the AI’s “objectivity” is actually just a reflection of the prejudices baked into its code. If the system was trained on historical data where innocent people were convicted, then THEMIS is just repeating those mistakes at lightning speed.
THEMIS pauses. For the first time in the movie, the AI hesitates.
The Verdict
THEMIS delivers its ruling: Guilty.
But then it adds something unexpected to the record: “Confidence level: 73%. Margin of error: unacceptable for a life sentence.”
The AI essentially admits it’s not 100% sure. Alex is convicted, but the system flags the case for human review.
What It All Means
The ending isn’t about whether Alex is innocent or guilty. It’s about whether we should trust machines to make life-or-death decisions.
THEMIS was supposed to be the perfect judge—no bias, no emotion, just facts. But the movie argues that “just facts” isn’t enough. Justice requires empathy, context, and the ability to admit uncertainty.
The final shot shows THEMIS processing millions of other cases, each decided in seconds. The implication? How many others were wrongly convicted because the AI was only “73% confident”?
Chilling.
Drop your theories below. Do you think Alex did it? 👇
Enjoyed this? Share it with a fellow movie fan!