AI and Domain Knowledge : Implications of the Limits of Statistical Inference

Today's AI is certainly powerful. But reliant as most of these systems are on machine learning, they have fundamental limitations. And these limitations have implications for AI's development, application, and impact in the coming years. Our second technology briefing will describe machine learning's fundamental nature as a form of statistical inference, and explore how this constricts effective use of today's AI to certain kinds of problem domains. Through the lens of this limitation, it will then take a position in three debates over how AI will unfold. First, will AI's diffusion throughout the economy and society be led by firms with the most advanced AI technical capabilities? Second, will China's access to unmatchable volumes of data lead to a dominant position in global markets for AI tools? And third, is the creation of AI that can match or exceed human intelligence, or Artificial General Intelligence (AGI), a foreseeable circumstance? Or is it a possibility only in the distant future, if it is possible at all?