The new model Astra developed by OpenAI has recently shown impressive performance in the field of mathematics, sparking widespread discussion in the tech industry and on social media. However, renowned scholar Gary Marcus wrote an article pointing out that the public's excessive enthusiasm for Astra is a typical "fallacy of composition," overestimating its versatility.

Gary Marcus emphasized that the success of artificial intelligence in mathematics is not accidental, but because mathematics is well-suited for verification using symbolic tools, and correct synthetic data can be generated at extremely low cost. However, breakthroughs in specific fields cannot be easily generalized to all cognitive tasks. Open-ended problems in the real world are far more complex than mathematics and cannot be solved through simple simulations.

Additionally, since OpenAI has not yet released specific method details or evaluation benchmarks for Astra, the understanding of its true capabilities remains very limited. Several experts have warned that improvements in mathematical ability do not automatically eliminate model hallucinations, nor do they indicate the arrival of artificial general intelligence (AGI). The public and the industry should maintain rationality and restraint when evaluating such technological advances.