Intelligence Is Not Intuition (olshansky.info)

The piece argues that while we already have broadly useful "general" AI — LLMs, multimodal generators and agentic tool users — what’s missing is artificial intuition (AGi): the tacit, experience-driven judgment that lets a system know which tool or reasoning shortcut to apply in messy, novel situations. Technically, today’s systems excel at learned pattern recognition and prediction (supervised/unsupervised training, RL with reward functions, inference), but they lack the equivalent of the "10,000 hours" of embodied, real‑world practice that produces human-style intuition and reliable judgment. For practitioners and researchers this distinction matters: intelligence scales logic and capability, but intuition scales judgment, safety and trust. The implication is that progress toward “superintelligence” won’t automatically yield human-like intuition; closing that gap likely requires long-term, grounded interaction, continual learning, and new evaluation frameworks for tacit knowledge and judgment. In practice, users must build intuition about model strengths and failure modes, and researchers should prioritize mechanisms for real‑world experience, lifelong learning, grounding, and robust decision-making if we want systems that don’t just compute well but choose wisely.