Leadership Change in 3 Months: Is the US AI Testing Institution Short on Talent or Consensus?

A head of an AI testing institution resigning after just 3 months in office indicates more than just a career choice behind it.

Chris Fole leaving the US Center for AI Standards and Innovation came suddenly, but upon reflection, it was expected. This institution was born to resolve a core contradiction: the US wants to hold discourse power in AI standard-setting but doesn't want to do so through strict regulation. Thus, what an institution nominally responsible for testing and standard-setting actually does may not even be clear internally.

Comparing the two routes makes it clear.

One is the European AI Act route: legislate first, then set up institutions, with institutions having explicit enforcement powers and penalty standards. The other is the US voluntary framework route: set up institutions first, then find funding, relying on industry self-governance to drive standards. Fole's resignation is, to some extent, a setback for this voluntary route in practice. An institution without explicit enforcement authority, without an independent budget, and possibly still negotiating team size leaves extremely limited room for its head to operate.

Data shows that US investment in AI testing and standard-setting is completely disproportionate to its investment in AI R&D. According to the Stanford AI Index Report, total private investment in AI in the US exceeded $67 billion in 2023, but funds used for public testing and standard-setting are estimated to be less than one-thousandth of the former. This imbalance makes AI testing institutions more of a "symbolic" existence rather than true "gatekeepers."

The core variable in this track has never been the technology itself, but the game between political consensus and commercial interests. The attitude of US tech giants towards AI testing standards can be seen by voting with their feet. Some companies build their own testing systems, while others choose to wait and see. When even those inside the institution feel there is "no leverage," the departure of the head is almost inevitable.

Fole's departure also exposes another deep-seated issue: the talent dilemma in the US AI testing field. Professionals who truly understand AI testing and have public management experience are extremely scarce globally. And an institution that frequently changes leadership finds it hard to build professional reputation, let alone attract top talent. This is a vicious cycle: the more unstable it is, the harder it is to hire people; the harder it is to hire people, the more unstable the institution becomes.

In contrast, although China's AI standard-setting started late, the model of concentrating strength to accomplish big things shows efficiency advantages in the standard-setting phase. The AI standard system jointly promoted by the Ministry of Industry and Information Technology and the Standardization Administration has released dozens of national standards, covering basic algorithms, data annotation, security assessment, and multiple links. This "top-down" promotion, while potentially lacking flexibility, at least guarantees execution and stability.

An open question for everyone: While US AI testing institutions are still fighting internally over "who manages" and "what to manage," will the global discourse power in AI standard-setting inadvertently shift to other regions?

Original link: https://www.ithome.com/0/979/288.htm