top of page
Search

How Should We Define Artificial Intelligence (AI)?

November 17th, 2023


The article discusses the significance of defining terms, particularly in the context of artificial intelligence (AI) and policymaking. 


Defining terms holds a profound significance, reminiscent of the ancient notion of a "true name" that encapsulates the essential nature of a concept. This act, similar to the magical art of naming, is mirrored in the process of writing legal definitions. The law breathes meaning into reality through words. In the realm of artificial intelligence (AI), the imprecision and inconsistency within policy discourse are notable. Discussions fluctuate between futuristic concerns about AI risks and descriptions that encompass even basic digital algorithms. The absence of a standardized international definition for AI compounds the challenge.

Policymakers struggle with the task of differentiating between high-risk and low-risk AI systems. While risk assessments are considered essential for many AI applications, more straightforward safeguards, such as third-party accountability mechanisms, are deemed necessary only for high-risk systems. Defining and designating high-risk systems present hurdles for policymakers, exemplified by debates around the inclusion of generative AI models in the EU AI Act. Ongoing discussions explore concepts like "horizontal exemption conditions" that might exclude certain AI systems from high-risk categorization based on limited purpose-based exemptions.


The U.S. Executive Order 14110 introduces a notable definition for AI, referencing the National AI Initiative Act of 2020. This definition outlines AI as a machine-based system capable of making predictions, recommendations, or decisions, influencing real or virtual environments based on human-defined objectives. The executive order further introduces complex definitions for high-risk AI systems, notably the "dual-use foundation model," which must perform tasks across multiple high-risk contexts. Clear definitions are crucial, as underscored by a bipartisan bill on AI accountability and transparency introduced by Senators John Thune and Amy Klobuchar, emphasizing the need for a precise and consistent understanding of AI in policymaking. The ongoing effort to refine policy discourse reflects the vital role that well-defined terms play in shaping the landscape of AI governance.



bottom of page