Did the ChatGPT Erdos controversy obscure a real achievement? (twitter.com)

The piece probes whether a noisy controversy over claims that ChatGPT had an low “Erdős number” — a playful metric of collaborative distance in math coauthorship networks — distracted from a substantive accomplishment: language models are increasingly able to contribute nontrivial, research-relevant artifacts. The flap centered on attribution and the absurdity of treating a trained model as an academic collaborator, which inflamed debates about authorship, citation, and whether model-generated content should be treated as intellectual contribution or mere synthesis of training data. The article contends that those procedural and ethical arguments overshadowed evidence that generative AI can accelerate problem exploration and produce verifiable, novel derivations when guided by human researchers. For the AI/ML community this matters because it highlights concrete technical and policy gaps: provenance of training data, evaluation methods that distinguish true novelty from regurgitation, and graph-based measures (like Erdős distances) that can be gamed by dataset leakage or superficial textual overlap. Practically, the incident underscores the need for standardized attribution metadata, benchmarks for verifying model-generated proofs or experiments, and clearer norms around coauthorship versus tool-use. The upshot: instead of fixating on catchy headlines, researchers should build tooling and standards that measure and validate AI contributions robustly, while addressing legal and ethical frameworks for credit and responsibility.