Active Learning and Causal AI Challenge the Claim That Only Humans Can Run the Theory-Experiment Loop

maier_ak · x · 2026-09-21

Andreas Maier critiques Felin & Holweg's position paper "Theory Is All You Need" (Strategy Science), which argues that LLMs are fundamentally backward-looking predictors, unlike humans' forward-looking, theory-driven causal reasoning that generates genuine novelty—a point illustrated via heavier-than-air flight. Maier concedes the argument is well-made but says it targets AI as of 2023: active learning, reinforcement learning, and causal models already design experiments, intervene, and create new data, undermining the claim that only humans can run the theory-experiment loop.

Related event: Researchers Rebut "Theory Is All You Need": AI Can Already Design Experiments(2 posts)→

Original post →

More from AGI Musings

AGI Musings channel →