AI Security Institute published a report on May 21, 2026, assessing oversight mechanisms for advanced AI systems against rapid capability gains and identifying degradation pathways.
Report section records disputed advantages of discrete token reasoning.
#20Miles Brundage@MILES_BRUNDAGE"Concerning" but unironically
AI Security Institute@AISecurityInstThe safety of advanced AI systems increasingly depends on the ability to oversee them. Our new report examines today’s AI oversight landscape, finding many pathways likely to lead to its degradation.🧵
From https://www.aisi.gov.uk/blog/will-it-become-harder-to-oversee-ai-systems
Miles Brundage@Miles_BrundageEncyclical or actual AI safety report, who is to say
Gotta click on the pic, bad cropping
Miles Brundage@Miles_BrundageFrom https://www.aisi.gov.uk/blog/will-it-become-harder-to-oversee-ai-systems
#1178Tomek Korbak@TOMEKKORBAKanother banger from UK AISI!
AI Security Institute@AISecurityInstThe safety of advanced AI systems increasingly depends on the ability to oversee them. Our new report examines today’s AI oversight landscape, finding many pathways likely to lead to its degradation.🧵
#13801a3orn@1A3ORNKudos to the one (??) expert in the report who pointed out that discrete token reasoning has better error correction, which is a factor decreasing the advantage of recurrent neuralese.
Also kudos to the report for tagging this as disputed.
AI Security Institute@AISecurityInstThe safety of advanced AI systems increasingly depends on the ability to oversee them. Our new report examines today’s AI oversight landscape, finding many pathways likely to lead to its degradation.🧵
@1a3orn is this a theoretical argument?
1a3orn@1a3ornKudos to the one (??) expert in the report who pointed out that discrete token reasoning has better error correction, which is a factor decreasing the advantage of recurrent neuralese. Also kudos to the report for tagging this as disputed.