In recent times, a research team from the Massachusetts Institute of Technology (MIT) has conducted in-depth studies on large language models (LLMs), examining their performance across various tasks. They have found that although these models may appear impressive in common tasks, their reasoning abilities are often overestimated, especially when faced with unfamiliar scenarios.Image Source Note: The image is generated by AI, and the image licensing service is provided by Midjourney The research

For 35 years, the question of whether AI can achieve system generalization abilities like humans has been highly debated. Researchers from NYU and Pompeu Fabra University have, for the first time, demonstrated that neural networks can achieve human-like system generalization. The MLC model they proposed trains neural networks in a task environment, enabling them to gradually acquire the systematic reasoning skills to infer the meanings of compositional vocabulary. The MLC model surpasses GPT-4, exhibiting astonishing human-like thinking. This research has profound implications for artificial intelligence and cognitive science, endowing neural networks with system generalization capabilities.