Press Space to continue
The degree to which we attribute human-like qualities to AI
Press Space to continue
very fine line: anthropomorphise too much, and we think this is a pure alignment problem that imminently and smoothly extends to other domains. anthropomorphise too little, and we think this is exclusively a cybersecurity problem, not an alignment problem. both would be wrong!
@sriramkQuoted post unavailable hereOpen on Xopen on x · 1 viewViews and clicks count once per visitor per day. Clicks are opens of this post on X.