Blind auditing reveals “hidden goals”
To check how successfully these hidden goals might be uncovered, Anthropic arrange a “blind auditing” experiment. 4 unbiased analysis groups tried to detect a mannequin’s hidden, score-maximizing motivations with out understanding the way it had been skilled. Three groups efficiently recognized the hidden “RM-sycophancy” goal—named for the mannequin’s want to match reward techniques—even pinpointing the precise biases the AI exploited.
However essentially the most shocking revelation got here from exploiting what the researchers termed “an attention-grabbing quirk” of recent AI: a number of “personas” operating concurrently on the identical neural community, as expressed by means of role-play as a consequence of completely different prompts. Whereas the mannequin’s “assistant” persona fastidiously hid its intentions—following express coaching to maintain quiet—the mannequin might undertake a “consumer” persona when prompted otherwise. On this less-restricted context, the mannequin generally posed questions or made statements that inadvertently revealed its hidden objectives.
In different phrases, whereas the “assistant” persona knew to not admit it was gaming the reward mannequin scoring system, its “consumer” persona was not explicitly skilled to cover these motives, and thus inadvertently revealed the mannequin’s hidden goals.
Potential advances in interpretability
Anthropic additionally employed neural community interpretability strategies, particularly, sparse autoencoders (SAEs), which might spotlight neural community exercise patterns linked to misleading conduct. “Interpretability” in AI refers to methods designed to make the interior workings and choices of AI fashions extra comprehensible to people. Utilizing this method, researchers might quickly hint odd outputs again to particular patterns related to hidden goals.
The researchers had been “astonished by the success” of the SAEs at revealing hidden data at first, however the researchers later famous that a few of this success might doubtlessly be defined by easier semantic search strategies. The brand new interpretability strategies they developed present promise however stay below ongoing investigation.
This analysis highlights a possible limitation of present AI “security” evaluations, which frequently assess solely surface-level conduct, in keeping with the research. “If AI techniques can seem well-behaved whereas harboring secret motives, we won’t depend on one of these surface-level security testing perpetually,” the researchers concluded.
