AI Applied to International Law

Artificial Intelligence (AI) Applied to Chess

Twenty-one years ago today, on February 10, 1996, an IBM supercomputer called Deep Blue beat then-reigning world champion Garry Kasparov in a game of chess, although Kasparov won the six-game match. In a rematch 14 months later, Deep Blue won the match.

chess set
photo by Alan Light
AI Applied to International Law

To win at chess, a computer has to predict how its human opponent is likely to move in response to moves the computer makes. So can a computer today predict how a human tribunal is likely to rule in a case? The answer may be yes.

Researchers have now created a machine-learning algorithm that has predicted the outcome of cases decided by the European Court of Human Rights with 79% accuracy. The algorithm was given text from the publicly available HUDOC database—summaries of submissions to the court. Based on language used, and topics and circumstances mentioned in the summary of each party’s submission, the algorithm predicted what the court would decide.

Would the results have been as accurate if the algorithm had been given the original submissions to the court, rather than the summaries? Unfortunately, the originals are not publicly available, so the answer is unknown.

Courtroom - European Court of Human Rights
photo by Adrian Grycuk CC BY-SA 3.0 pl
More on AI and the Law