The AI folks do not seem to understand intelligence

The article challenges the "scale is all you need" mantra driving AI development. OpenAI's rapid progress on complex math problems does not guarantee universal intelligence. Intelligence requires constant, clear feedback to remain effective. Without it, models can drift from reality, similar to Samuel Langley's failed 1903 flight attempt. His well-funded, stable design still plunged into the Potomac River.

Asia's tech sector should note where AI excels: coding, mathematics, and weather forecasting. These fields offer vast data and rapid feedback. Financial markets, like those served by Two Sigma Investments with its top-five supercomputing power, also fit this model. This means Asian firms in these sectors will see the fastest and most profound AI integration. Other areas, like drug discovery, will face slower adoption.

The real test for Asian AI companies is identifying problem sets with clear, fast feedback loops. This is where intelligence delivers maximum impact. Expect continued investment in AI for financial services and software development across the region. However, the returns on AI in areas like personalized content creation or complex scientific research will diminish. This is because the feedback is slower and less precise.