Regulating AI: Rewrite the Rules First, or Rename It?

Two distinct approaches to AI governance have emerged recently: one starts with mechanism design to make AI models "willingly" follow rules, while the other centers on renaming, rebranding AI as "Super Intelligence" to see who adopts the new term. Both approaches are tangible, but neither has resolved the core challenges yet.

Reverse Game Theory: Make AI Voluntarily Follow the Rules

Amit Katwala from MIT Technology Review introduces a framework called "reverse game theory". Traditional game theory predicts how rational participants will act under existing rules; reverse game theory flips this logic to design rules that intentionally elicit desired behaviors. The original article uses the classic cake-splitting analogy: if one person cuts the cake and the other picks their slice first, the cutter has a clear incentive to split the two portions perfectly evenly.

Researchers argue this method can make AI far more aligned with the rules and instructions we set for it. But the problem is very straightforward: there is still no evidence that these mechanisms work in real-world practice. Elegant as mechanism design sounds, it may not follow the intended script when applied to AI models that are adept at exploiting loopholes. Original article

Is Renaming a Form of Governance?

At the end of September, Trump signed an executive order with only one mandate: all government agencies must rebrand artificial intelligence as "Super Intelligence". A month later, the new term has slowly started appearing on government websites and official documents, though usage remains inconsistent. Many people at major tech giants still continue to use the term "AI".

A bigger concern is that the new term is turning into a political loyalty test. People who use the term are effectively demonstrating allegiance to the administration, while those who refuse may become targets for retaliation. James O'Donnell asks in his original piece: one month on, who has fallen in line? Renaming is far cheaper than revamping entire regulatory systems, and it makes it much easier to tell which side people are on. Original article

Today's Must-Read Roundup

• The US midterm elections hold polls today, with candidates betting that AI regulation stances will sway voters. Per Axios reporting, tech investors are on tenterhooks over the outcome.
• Nvidia is requiring its data center clients to limit their China-facing operations, mandating that most of their sales come from non-Chinese customers. Both FT and Bloomberg attribute this rule to pressure from Washington.
• Cyberattacks targeting connected supermarket refrigeration systems have hit Europe, prompting US grocery store networks to go on high alert. As refrigerators gain internet connectivity, food safety now has a new cyber attack entry point.
• CERN's Large Hadron Collider has detected a signal at the smallest scale ever observed, which could provide clues to undiscovered new particles.
• Stimulating a specific region of the brain can make people put more cognitive effort into thinking, but reduces their motivation to do so. Performance and willingness are effectively decoupled in this scenario.
• Big tech firms are building massive natural gas power plants to support their data centers, putting their public climate commitments at risk of being rolled back.
• Google states that one of its Android partners got compromised, infecting 1.3 million devices, mostly Android TV boxes and similar consumer hardware.
• Apple is working on its first consumer robot, and the project is proving far harder than expected. Useful, practical AI robots are still a long way off.
• During China's Double 11 shopping festival, AI tokens have emerged and been sold as an entirely new product category.
• A Sahara meteorite brings new clues about the history of water on Mars.

Today's Quote

"Don't waste your money paying human experts to debate robot experts," family lawyer Steven Benmor told the Wall Street Journal. The comment comes in context of clients bringing AI chatbot-generated divorce advice to law firms, where lawyers have to charge fees to correct the bad guidance. The consultation fees people save by using AI sometimes end up being paid out in far more expensive ways later.

Human-Caused Noise Is Forcing Animals to Change Their Calls

Clive Thompson writes about an often-overlooked consequence of human societal expansion: animals are finding it increasingly hard to hear each other over human-generated noise. Many bird species are forced to sing faster with denser trills to be heard, which makes their mating calls far less effective. Noise pollution also increases inter-bird conflicts, while species that cannot tolerate urban loudness simply abandon the areas entirely.

New technical solutions are being developed to cut down this noise pollution, which could also benefit human communities. Original article

Something Light to Wrap Up With

Smithsonian's live stream of cheetah cubs; Stevie Wonder releasing four unreleased lost tracks from Songs in the Key of Life; Finland's 30th Air Guitar World Championships; winning entries from the Ocean Photographer of the Year award. These are the fun, lighthearted highlights in today's brief.