Are we on the cusp of true artificial intelligence, or are we being sold a cleverly disguised illusion? This episode examines a critical report from Apple researchers that casts serious doubt on the reasoning capabilities of Large Language Models (LLMs) like ChatGPT.The episode explores evidence presented by Apple researchers suggesting that LLMs, despite their impressive feats, may primarily rely on memorization and pattern matching rather than genuine reasoning. The researchers tested popular LLMs using modified datasets designed to expose their limitations. They found that even slight changes in wording or the introduction of irrelevant information could dramatically impact the models’ performance, suggesting a lack of deep understanding.