Karen Hao on the Empire of AI, AGI evangelists, and the cost of belief (techcrunch.com)

Journalist Karen Hao argues in Empire of AI (and in a recent TechCrunch interview) that the AI industry has developed an ideology — a quasi-religious commitment to building artificial general intelligence (AGI) — with OpenAI as its chief evangelist. That belief system prizes speed and scale above efficiency, safety, and diverse research paths, driving companies to “pump more data, more supercomputers” into existing techniques. Hao warns this empire-like consolidation has rewired geopolitics and research priorities: top talent has migrated from academia to industry, billions are being poured into compute (OpenAI projects ~$115B burn by 2029; Meta and Google have similarly massive AI capex plans), and products are released rapidly even as harms accumulate. Technically and ethically, Hao contrasts two approaches: the scale-driven LLM/AGI trajectory (large scraped datasets, huge energy and infrastructure costs, and risks like hallucinations, job displacement, concentrated wealth, and toxic-content exposure for low-paid workers) versus algorithmic and domain-specific systems that deliver concrete benefits with less cost — e.g., DeepMind’s AlphaFold, which uses targeted biological data to solve protein folding without the same social and environmental harms. She also flags structural conflicts at OpenAI (hybrid nonprofit/for-profit model, Microsoft ties, possible IPO) that may blur “benefiting humanity” claims and incentivize growth over accountability.