LESSWRONGTags
LW

AI Boxing (Containment)

•

Applied to Disproving and partially fixing a fully homomorphic encryption scheme with perfect secrecy by Lysandre Terrisse 6d ago

•

Applied to An AI, a box, and a threat by jwfiredragon 3mo ago

•

Applied to The case for training frontier AIs on Sumerian-only corpus by Charbel-Raphaël 5mo ago

•

Applied to Why do so many think deception in AI is important? by Gunnar_Zarncke 5mo ago

•

Applied to Planning to build a cryptographic box with perfect secrecy by Lysandre Terrisse 5mo ago

•

Applied to Protecting against sudden capability jumps during training by nikola 6mo ago

•

Applied to Information-Theoretic Boxing of Superintelligences by JustinShovelain 6mo ago

•

Applied to Self-shutdown AI by jan betley 9mo ago

•

Applied to Boxing by Raemon 10mo ago

•

Applied to Thoughts on “Process-Based Supervision” by Steven Byrnes 11mo ago

•

Applied to A way to make solving alignment 10.000 times easier. The shorter case for a massive open source simbox project. by AlexFromSafeTransition 1y ago

•

Applied to [FICTION] Unboxing Elysium: An AI'S Escape by Super AGI 1y ago

•

Applied to Ideas for studies on AGI risk by dr_s 1y ago

•

Applied to How to safely use an optimizer by Simon Fischer 1y ago

•

Applied to ChatGPT getting out of the box by qbolec 1y ago

•

Applied to ARC tests to see if GPT-4 can escape human control; GPT-4 failed to do so by Christopher King 1y ago

•

Applied to Bing finding ways to bypass Microsoft's filters without being asked. Is it reproducible? by Christopher King 1y ago