Critic Highlights Declining Transparency In AI Amid Rising Stakes

I truly miss the age of science and transparency in AI. Especially given how much money and political power and governance and scientific understanding is at stake. We don’t know for example • How many Erdos problems OpenAI tried and succeeded vs failed on. • Whether GPT-5.5 (which can also find the counterexample) was trained on the newly-discovered counterexample. • How if at all the new model differs from the old model • What it was trained on • How much compute was used in the search for a compelling example. • How well the new system does on traditional benchmarks, or on problems that are less formalizable The answers matter, but all live behind closed doors.