The newsletter reports that a new approach called "reverse game theory" aims to control powerful AI models by designing rules that incentivize them to follow instructions. Standard game theory predicts how rational actors behave under given rules, while reverse game theory designs rules to produce a desired behavior. The source offers a cake-splitting analogy: if one person cuts and the other chooses first, the cutter has reason to make equal portions. Researchers believe the approach could align AI more closely with human rules and instructions, but the source says big questions remain about whether these mechanisms will work in practice.
The source also reports that in late September, President Trump signed an executive order directing government agencies to rebrand artificial intelligence as "Super Intelligence." According to the source, the new term is slowly appearing on government websites and documents, though some major tech companies still use "AI." The source says a concern is that the term is becoming a political loyalty test, with people who use it seen as signaling allegiance to the administration and those who don't becoming targets for attacks.
The source also notes that today's US midterm elections will test AI's influence on American politics, and that candidates are betting AI regulation will sway voters.