Don't Stop Believin' in OpenAI (buildcognitiveresonance.substack.com)

OpenAI’s hype cycle is under scrutiny after months of contradictions between grandiose claims of near-term AGI and a modest product release. The company’s CEO framed AGI and superintelligence as near-solved problems, even as reporters and insiders flagged training hiccups for the next flagship. When OpenAI shipped ChatGPT‑5 in August, the model showed only incremental gains rather than the leap many expected — a sign that simply adding more data and compute didn’t deliver the promised qualitative jump. The rollout included puzzling marketing artifacts (an AI‑generated “graph”) that reinforced skepticism about whether the technical story matches the corporate narrative. The situation matters because it mixes technical plateauing with a financial and political feedback loop: mega‑deals, partner commitments (e.g., chip and infrastructure purchases), and market cap boosts are being driven as much by belief in a future product as by current capabilities. That creates systemic risk — a bubble-like structure akin to Enron-style belief-driven finance — with implications for research funding, hardware supply chains, evaluation standards, and regulation. For the AI/ML community the lesson is clear: demand stronger empirical transparency, realistic benchmarking, and cautious funding assumptions. The author adds a civic note — large protests planned across the U.S. on October 18 — underscoring how the debate has spilled from labs and boardrooms into public activism.