The reason enterprise AI is stuck (www.fastcompany.com)

Enterprise AI is currently stagnating, not due to weak models or short context windows, but because the industry relies heavily on human metaphors to conceptualize AI systems. These metaphors—like memory, reflection, and even "dreaming"—help simplify complex technologies for product teams and executives alike, making them more relatable. However, this approach hinders the industrialization of AI because metaphors describe rather than formalize the underlying computational architectures. This reliance on metaphorical interpretations creates a disconnect between the impressive capabilities demonstrated in AI demos and the frustrating realities of deployment in enterprise settings. As companies struggle to move beyond abstract representations to more concrete models, the potential for wide-scale adoption and optimization of AI solutions remains hampered. Recognizing this distinction could lead to more effective strategies for developing and implementing AI technologies in a way that truly leverages their capabilities.