Physical Superintelligence says AI system proved quantum conjecture; Meta details sensing glove
Reported by Crypto Briefing and Import AI2 sources ↓
Physical Superintelligence says a system using a non-public Anthropic model produced a Lean-checked proof of a long-open quantum conjecture. Import AI also reports on Meta’s Tactus glove and a NASA study of lunar manufacturing costs.
Key points
- PSI says its system completed the proof in three days and improved the result twice
- The proof’s Lean formalization is public, but the model is unreleased and independent review is underway
- Meta describes Tactus, a textile glove for recording hand pose and touch during manipulation
- A NASA study estimated lunar manufacturing could start with 12–41 metric tons of launched material
Sources
Details
Physical Superintelligence announced the claimed proof on November 6, according to Crypto Briefing. The Cambridge lab says its research system, combining proprietary infrastructure with an unnamed, non-public Anthropic model, produced a complete proof of the Aaronson-Ambainis conjecture in three days; Lean accepted the formalization. PSI says the system then improved the result twice.
Crypto Briefing reports that PSI published a paper attributed to David Turturean and released a 9,454-line Lean formalization. The outlet says the Stanford researchers who proved a special case in August 2026 are among those reviewing the claim; independent verification is still underway. A proof checker can verify the formalized theorem, but not whether that formalization accurately represents the original conjecture.
Import AI reports that Meta Reality Labs Research published details of Tactus, a textile-integrated glove designed to record tactile and hand-pose data during object handling. Meta describes a 1.5-millimeter fingertip stack, strategically placed capacitive sensors and pose tracking that uses inertial measurement units to maintain accuracy when hands are occluded by objects.
Import AI also summarizes a 2016 NASA study of lunar manufacturing. The study’s baseline case called for launching 41 metric tons of material, while a smaller-scale case used 12 metric tons and delayed the industry’s benefits by eight years. Import AI estimates that delivering those amounts to the Moon could cost $1.5 billion to $4.2 billion at list prices, assuming multiple Starship launches and in-orbit refueling.
Written from coverage as of
Images
- Crypto Briefing
Stocks in this story
Change since open: 20 Nov to 21 Nov 2026
Last refresh