Gpt 5,6,7: Does it even matter? The (ghost) productivity question. [D]

Despite the genuine capabilities of current AI models like GPT-5, which can perform a substantial fraction of knowledge work, a noticeable productivity shock in the real economy has not yet occurred. These systems excel at tasks such as writing, summarizing, analyzing documents, explaining technical concepts, generating code, and conducting research, previously requiring educated human labor. The question remains whether AI is less economically useful than benchmarks suggest, or if organizations are too slow to translate model capabilities into measurable output.