Report claims AI technology suffers “complete accuracy collapse”

Apple’s latest research paper reveals that advanced AI models, particularly large reasoning systems, suffer a “complete accuracy collapse” when faced with complex challenges, while even large language models struggle with sophisticated tasks. The study, conducted by Apple researchers and published in a technical paper, found that these models reduce their reasoning effort as they encounter difficulty, raising concerns about their reliability. Associate Professor Niusha Shafiabady of Australian Catholic University warned that expecting AI to be a magic wand is a mistake, emphasizing the limits of current approaches. AI critic Gary Marcus described the findings as “pretty devastating” and criticized the hype surrounding AI as a “giant game of bait and switch.” The report also highlighted real‑world errors, such as Apple’s AI‑powered news alerts misreporting high‑profile events, and cited additional inaccuracies from other major AI assistants, underscoring the broader issue of AI reliability.