USA: "Super Intelligence" Mandated, Incentives to Regulate AI

Key Takeaways

1At the end of September, a decree mandates federal agencies to use the term "Super Intelligence" to refer to AI.

2A month later, adoption remains partial, and major players continue to use "AI."

3Researchers are promoting the theory of reverse games to align AI agents, with uncertainties about its effectiveness.

Why it matters : These developments highlight a political framework and competing technical approaches to regulate the behavior of powerful AI agents.

In the United States, AI is caught between a strong political framework and technical alignment proposals. A decree signed at the end of September mandates the use of the term "Super Intelligence" by federal agencies, while researchers advocate for incentive systems to contain the behavior of powerful agents. A month later, the adoption of the new term remains incomplete, and the effectiveness of the proposed mechanisms remains uncertain.

In the United States, AI Influences the Electoral and Ideological Battle

The regulation of AI has become a divisive political issue in the United States. The country remains split on how to regulate AI. Some candidates are betting on the impact of AI regulation to persuade voters, while tech investors express their concerns as election results approach. Discontented voters could also challenge Republican bets related to Elon Musk. Additionally, tensions surrounding Chinese AI have reached the White House.

Renaming AI to "Super Intelligence": A Directive, Incomplete Adoption

At the end of September, a presidential decree requested that U.S. government agencies use the term "Super Intelligence" instead of "AI." A month later, this new term is starting to appear on some websites and official documents, but it is not yet widespread. Several major tech companies continue to use "AI." Concerns exist that the adoption of the term could serve as a test of political loyalty: using it might signal allegiance to the administration, while not using it could expose one to attacks.

Le brief IA que lisent les pros

Le brief IA que les pros lisent chaque soir

Les 7 actus IA du jour, décryptées en 5 min. Gratuit.

Inclus dès l'inscription : notre sélection des meilleurs guides & comparatifs IA.

Choisissez votre rythme

Gratuit · Pas de spam · Désabonnement en 1 clic

Incentive Mechanisms to Contain AI Agents

Beyond the vocabulary issue, researchers propose a technical approach to frame powerful AI models: reverse game theory. While classical game theory is used to anticipate the behavior of rational actors under rules, the reverse version involves designing the rules themselves to achieve the desired behavior. The example of sharing a cake illustrates this principle: if one person cuts and the other chooses, the cutter has an incentive to make equal shares. According to researchers, this method could allow for more effective alignment of AI agents with human instructions, although doubts remain about the actual effectiveness of these mechanisms in practice.

What Users Are Already Experiencing: Paying to Correct Chatbots

In certain sectors, the concrete effects of AI are already being felt. Steven Benmor, a family law attorney, notes that clients increasingly need to consult human experts to correct the advice given by chatbots, resulting in additional costs.