The Machines Have Won: AI Lessons for E-Discovery from the World of Chess

In 1997, IBM’s Deep Blue defeated the then world chess champion, marking a symbolic passing of the torch from humans to machines. Since then, the gap between human and computer performance has only widened. Today, even mobile chess apps like Stockfish are capable of defeating the best human players in the world.

While no serious chess player disputes that AI has surpassed humans in chess, many legal professionals still prefer “human eyes” when reviewing electronically stored information (ESI). In a recent Exterro poll, 47% of respondents said they would rather rely on attorneys than AI for document review. However, this perspective is increasingly outdated.

A 2011 academic study on technology-assisted review (TAR) by Maura Grossman and Gordon Cormack found that:

“Technology-assisted processes, while more efficient, can also yield results superior to those of exhaustive manual review, as measured by recall and precision.”

Modern AI Is Radically Different from Its Ancestors

When Deep Blue defeated Garry Kasparov, it relied heavily on brute-force computation—evaluating hundreds of millions of moves per second—combined with input from human chess experts.

Modern AI operates very differently.

Systems like DeepMind’s AlphaZero learn independently, using only the rules of the game as input. AlphaZero taught itself chess in under four hours and then defeated Stockfish 8—the world’s leading chess engine at the time—in a 100-game match without losing a single game, despite evaluating far fewer positions per second.

In e-discovery, earlier machine learning methods like predictive coding depend on human-created “seed sets” to classify documents. While effective, these methods are limited by their initial human input.

By contrast, newer deep learning systems can learn continuously by observing human reviewers and refining their predictions over time. This iterative, unsupervised learning allows AI to improve beyond the limitations of its initial training.

Human-AI Collaboration Is More Powerful Than Either Alone

After Deep Blue’s victory, the idea of humans competing against machines in chess became less meaningful. However, a new format emerged: freestyle chess, where humans and AI work together.

Interestingly, the winners of these competitions were often not the strongest human players or the most powerful machines, but teams that combined human insight with AI capabilities and effective processes.

Garry Kasparov himself observed:

“Weak human + machine + better process was superior to a strong computer alone—and even superior to a strong human + machine + inferior process.”

What This Means for E-Discovery

The takeaway for e-discovery is clear: AI is not replacing humans—it is augmenting them.

Given the massive volume of data involved in modern litigation, relying solely on human review is no longer practical. At the same time, AI alone is not sufficient. The best outcomes come from effective collaboration between legal professionals and intelligent systems.

When used correctly, AI can enhance accuracy, reduce costs, and accelerate timelines—while human expertise ensures context, judgment, and strategic decision-making.

Artificial intelligence has already become a major force in the legal industry, and its influence continues to grow. Understanding how to effectively integrate AI into e-discovery workflows will be critical for legal teams moving forward.