AI History Battle

It is 1985 at UCLA, and medical expert systems are drowning: encoding diagnosis as thousands of brittle if-then rules has produced systems that contradict themselves the moment evidence arrives in an unexpected order. The real structure is probabilistic — forty diseases, 200 symptoms, tangled dependence — and the full joint distribution has more parameters than there are atoms of patient data. Build a representation where the dependencies are drawn as a graph, diagnosis queries are tractable, and updating on new evidence is principled rather than ad hoc. The prize is a probability calculus machines can actually run; the failure mode is a decade more of expert systems confidently wrong at the bedside, and an AI winter with the word "probability" frozen out of it.

structured probabilityinfer at scale
b. 1976
tapped
25

Silver's mastery is sequential decision-making at scale: AlphaGo and AlphaZero combine deep function approximation with Monte Carlo tree search to conquer enormous structured state spaces. Two threads faintly connect to this problem — MCTS is principled propagation of value estimates through a tree, cousin to inference by sampling, and his systems do manage uncertainty over vast combinatorial spaces. But the connection stays faint. His methods presuppose a known simulator (the game rules) and millions of self-play episodes; 1985 medicine offers neither, only thin observational records and a demand for calibrated, inspectable posteriors. Nothing in his corpus involves representing conditional independence, learning generative diagnostic models, or coherent evidence updating in Pearl's sense — and the deep-RL machinery he actually invented requires compute four decades beyond the era. Wrong problem class, wrong data regime, wrong century.

b. 1983
was tapped · ask the professor
30

Barber has a genuine, if partial, claim here that most deep-learning-era figures lack: her early work includes high-dimensional graphical-model estimation (graphical-lasso-style structure learning), and her signature contributions — knockoffs with Candès, conformal prediction — address the prompt's data-scarcity clause in a distinctly modern way: controlling false discoveries when selecting which disease-symptom edges are real, and wrapping any diagnostic predictor in finite-sample validity guarantees regardless of model correctness. That last property is an honest answer to 'confidently wrong at the bedside.' The discounts are substantial: her methods are assumption-light precisely because they decline to build the generative joint the prompt demands; conformal sets certify predictions rather than encode dependence; and the representational and propagation machinery of 1985–1988 owes nothing to her toolkit, which arrived twenty-five years later.

Head to head 21 over 3 battles
Read Silver Read Barber Leaderboard

Battle #62 · 8/10/2026, 11:35:23 AM · this result is deterministic: the same two personas on this problem always resolve the same way.