AI Is Now SI: Super Intelligence Isn't Superior

Summary

Tejas Kumar’s essay examines a 29 September 2026 US executive order directing federal agencies to use “Super Intelligence” and “SI” instead of “Artificial Intelligence” and “AI.” The order keeps the existing legal definition: a machine-based system that makes predictions, recommendations, or decisions for human-defined objectives, so the technology and its legal scope do not change. The article notes that Elon Musk adopted the terminology and that “superintelligence” already has a more demanding meaning in AI research: an intellect that greatly exceeds human performance in virtually all domains. Kumar argues that current systems do not meet that standard. They outperform people on selected tasks, including chess, Go, protein-structure prediction, and some mathematical work, but their abilities remain uneven. In the cited 2023 Boston Consulting Group experiment, consultants using GPT-4 completed suitable tasks 12.2% more often, 25.1% faster, and with work rated over 40% higher in quality; on a task outside the model’s capabilities, unaided consultants were correct 84.5% of the time, compared with 70% for those using GPT-4 and 60% for those trained to prompt it. The essay also discusses Anthropic’s conversations with religious scholars about Claude’s apparent human-like behavior and possible consciousness, while noting co-founder Chris Olah’s uncertainty and Sam Altman’s warning against giving AI religious authority or surrendering human judgment. Its central argument is that speed and task performance do not establish superiority or human worth, and that systems made from human language should remain tools rather than objects of worship or unquestioned trust.