Calling or referring to Artificial Intelligence (AI) as a single technology is convenient but misleading.
AI includes a family of related technologies:
- predictive machine learning;
- language and generative models;
- computer vision and speech systems;
- recommendation and decision systems;
- agents and robotics;
- specialized scientific and industrial models.
These systems succeed and fail in different ways, using different architectures, data, interfaces, and evaluation methods. They also create different benefits, costs, and risks. A customer-service chatbot and an autonomous vehicle are both called AI, but they should not be evaluated or governed as though they were the same technology.
Yet the family shares a common foundation: computation, data, learned representations, adaptable models, and integration with tools and workflows. Together, these capabilities can spread across industries and stimulate complementary innovation.
AI is technically a family of related technologies but economically may function as a general-purpose technological platform.
This is similar to computing. Computing is not one machine or application; it is a family of hardware, software, networks, and methods whose combined diffusion transformed the economy.
Therefore, never evaluate AI in the abstract. Instead ask:
- Which AI system?
- Performing which task?
- Using what data?
- Under what conditions?
- With what risks and evidence of value?
It’s okay to use the singular when discussing AI’s economy-wide potential but we should use the plural when making real decisions.
Happy Building!
Leave a Reply