In a recent commentary in Nature, Alan Turing’s legacy becomes central as the authors claim that artificial general intelligence (AGI) is present in large language models (LLMs) due to their breadth and depth across multiple domains. This assertion stems from functionalist theory, which argues that if machines exhibit intelligent behavior, they are deemed intelligent. However, while LLMs can mimic human-like thought, they do not possess self-awareness or true understanding. The article asserts that the Turing Test merely signifies that a machine behaves as if intelligent, not that it comprehends. The discussion highlights the risks of misattributing intelligence to these systems, which merely shuffle language without real comprehension. This illusion can erode our ability to discern truth from fiction, underscoring a need for critical evaluation of AI’s role. Instead of seeking AGI, we should recognize machines for what they do and redefine our understanding of knowledge and trust in technology.
