Alexander Pan

5 papers · 2022–2024 · 3 venues · across top CS/AI venues

Authors are matched by name, so researchers who share a name ("Yang Liu", "Wei Wang") appear as a single profile and their paper counts, venues and achievements are combined. How identity is handled.

Achievements

Jump to papers ↓

+1 more ↓

🌍 Conference Polyglot · Bronze 🌉 Interdisciplinary Bridge · Bronze 🗺️ Taxonomy Completionist (13) 🐣 Hot Topic Early Bird 👥 Mega-Team · Squad

❓ The Questioner (2)

Venues

ICML (3) ICLR (1) NeurIPS (1)

Top co-authors

Steven Basart (3) Dan Hendrycks (3) Alice Gatti (2) Stephen Fitz (2) Nathaniel Li (2) Andy Zou (2) Mantas Mazeika (2) Jacob Steinhardt (2) Gabriel Mukobi (2) Adam Alfred Hunt (1)

Research topics

Machine Learning (4) Artificial Intelligence (3) Core AI (3) Natural Language Processing (2) Application Areas (2) Learning Types (2) Optimization & Theory · Machine Learning (2) Deep Learning (1) Applications (1) Core Methods (1) Optimization & Theory · Deep Learning (1) Resources & Methods (1)

Keywords

large language model (2) benchmark evaluation (1) anomaly detection (1) policy gradient (1) ai safety (1) spectral analysis (1) feedback loop (1) reward hacking (1) model scaling (1) safety benchmark (1) capabilities component (1) negative side effect (1) reward optimization (1) ethical behavior (1) machine ethics (1) policy refinement (1) proxy reward (1) model capabilities (1) output refinement (1) scale correlation (1)

Papers

Feedback Loops With Language Models Drive In-Context Reward Hacking ICML 2024 Safetywashing: Do AI Safety Benchmarks Actually Measure Safety Progress? NeurIPS 2024 The WMDP Benchmark: Measuring and Reducing Malicious Use with Unlearning ICML 2024 Do the Rewards Justify the Means? Measuring Trade-Offs Between Rewards and Ethical Behavior in the Machiavelli Benchmark ICML 2023 The Effects of Reward Misspecification: Mapping and Mitigating Misaligned Models ICLR 2022