Famed investor Michael Burry voiced strong criticism of the current trajectory of AI development, arguing that the industry is caught in a fundamentally flawed approach. In a Substack note, Burry introduced 'Ballard’s Test,' which posits that an entity cannot truly possess understanding unless it can demonstrate reason entirely in the absence of language. He noted that the AI field's original goal was to generate this true capacity for reasoning first, but because that proved too difficult, researchers abandoned it and took the easier path by pivoting to language-first models. Burry labeled this pivot a known flaw and a bad start, arguing that it has resulted in a parameter trap where the industry is stuck scaling up existing methods rather than innovating true cognition. He further highlighted the irony that this trap relies heavily on brute-force language processing powered by zillions of power-hungry chips, creating an incredibly ironic bottleneck for future advancement.