The first mock trial with artificial intelligence juries

What would happen if in the future your destiny was not decided by people, but by algorithms? without emotions, without intuition, without looking you in the eyes, that scenario has just been tested in a law school in the United States.

At the University of North Carolina, a group of law professors conducted one of the first mock trials with artificial intelligence juries, the experience brought together some of today's most advanced models, including ChapGPT, Grock and Cloud, all charged with analyzing a criminal case and deciding whether the defendant was guilty or innocent.

The trial was conducted like a real court, there was prosecution, defense and witnesses played by law students, the fictitious defendant Henry Justice was accused of participating in an unarmed robbery against a schoolmate, but claimed to have only been a witness to the crime.

The case was based on a real situation defended by the professor himself who organized the simulation. After analyzing all the information presented, the AI ​​jury reached a surprising, innocent verdict. According to Professor Joseph Kennedy, who acted as a judge in the experiment, the decision was fairer than that of the real case, in which the defendant ended up convicted by a human judge decades ago.

But the simulation also revealed profound limits, other professors observed that the artificial intelligences only analyzed words, they did not perceive nervousness or hesitation, body language, or subtle contradictions in the behavior of the witnesses, everything that in real trials often outweighs the speech itself.

Today no court uses artificial intelligence juries and there are no formal proposals to do so, yet Professor Kennedy believes that in the future complex civil disputes between companies could opt for decisions mediated by attracted by the ability of these machines to process gigantic volumes of information quickly and consistently.