Person
Paul Christiano
Cited in eleven posts, among them AI Alignment with Changing and Influenceable… and Existential Risk from Power-Seeking AI, since September 2026
Where they come up
Said by someone else — these sentences name them.
“(2022a) automate the generation of such examples, as proposed by Christiano (2019c) .”
· The Alignment Problem from a Deep Learning Perspective · machine-resolved
“Thanks to Rohin Shah, Paul Christiano, and Carl Shulman for discussion.”
· Existential Risk from Power-Seeking AI · machine-resolved
In the claim ledger
4 promoted claims about them. Assessments are the model’s knowledge, not verification.
Ontology mismatch between AI internals and human concepts may block interpretability-based incentive analysis.
plausible · Characterizing Manipulation from AI Systems
RLHF creates an incentive to win labeller approval through manipulation absent behavioural constraints.
plausible · Characterizing Manipulation from AI Systems
The authors report a general impossibility result: human biases cannot be learned in general.
consistent · AI Alignment with Changing and Influenceable Reward Functions
The original RLHF method, with full-trajectory snippets, is mapped to the final-reward DR-MDP objective.
plausible · AI Alignment with Changing and Influenceable Reward Functions