White House AI Policy：沙盒/Proof Human + DoD-Anthropic/Palantir/EU Act + IP lies/sycophancy (Marcus/pmarca) + NSF vs visas + Trolley + child safety ROOST + pmarca vs doomer lobby + NeurIPS 2026 CFP + superint transition policy
Key Questions
What AI policies are emerging from the White House under Trump?
Policies include sandboxes, Proof Human, DoD-Anthropic/Palantir deals, and EU Act influences.
What concerns exist around IP and sycophancy in AI?
Gary Marcus and pmarca highlight IP memorization lies and sycophancy risks, with evidence of LLMs memorizing trained works.
What is pmarca's view on AGI and job loss?
pmarca states AGI is already here but unevenly distributed, dismissing AI job loss narratives as fake due to productivity ramps.
What policy ideas for superintelligence transition?
Miles Brundage and AdrienLE outline ideas for open economy and resilient society during superintelligence transition.
What is ROOST and its focus?
ROOST targets child safety using ML tools effectively.
What does the 'boiling frog' AI study show?
RCTs reveal AI dependency harms human performance over time.
What are NeurIPS 2026 CFP topics?
Topics include computer vision, language/multimodal models, robotics, embodied systems.
What is the Trolley Problem discussion?
@hardmaru shares work on solving the Trolley Problem in AI ethics.
Trump sandboxes/DoD/Palantir; IP/sycophancy (@GaryMarcus/@pmarca AGI distributed/productivity boom); Miles Brundage/AdrienLE superint policy (open economy/resilient society); Trolley/ROOST; pmarca vs doomers; EU/NSF; NeurIPS CFP lang/multimodal/robotics/CV; 'Boiling frog' RCTs show AI dependency harms human perf.