Search Results for author:
Found 26 papers, 12 papers with code
Inference-Time-Compute: More Faithful? A Research Note
We refer to these models as Inference-Time-Compute (ITC) models.
The Two-Hop Curse: LLMs trained on A$\rightarrow$B, B$\rightarrow$C fail to learn A$\rightarrow$C
[Notice: This version is outdated.
Towards evaluations-based safety cases for AI scheming
We sketch how developers of frontier AI systems could construct a structured rationale -- a 'safety case' -- that an AI system is unlikely to cause catastrophic outcomes through scheming.
Looking Inward: Language Models Can Learn About Themselves by Introspection
We study introspection by finetuning LLMs to predict properties of their own behavior in hypothetical scenarios.
Me, Myself, and AI: The Situational Awareness Dataset (SAD) for LLMs
Chat models, which are finetuned to serve as AI assistants, outperform their corresponding base models on SAD but not on general knowledge tasks.
Connecting the Dots: LLMs can Infer and Verbalize Latent Structure from Disparate Training Data
As a step towards answering this question, we study inductive out-of-context reasoning (OOCR), a type of generalization in which LLMs infer latent information from evidence distributed across training documents and apply it to downstream tasks without in-context learning.
Can Language Models Explain Their Own Classification Behavior?
To explore this, we introduce a dataset, ArticulateRules, of few-shot text-based classification tasks generated by simple rules.
Tell, don't show: Declarative facts influence how LLMs generalize
Nevertheless, the effect of declarative statements on model likelihoods is small in absolute terms and increases surprisingly little with model size (i. e. from 330 million to 175 billion parameters).
How to Catch an AI Liar: Lie Detection in Black-Box LLMs by Asking Unrelated Questions
Large language models (LLMs) can "lie", which we define as outputting false statements despite "knowing" the truth in a demonstrable sense.
The Reversal Curse: LLMs trained on "A is B" fail to learn "B is A"
If a model is trained on a sentence of the form "A is B", it will not automatically generalize to the reverse direction "B is A".
Taken out of context: On measuring situational awareness in LLMs
At test time, we assess whether the model can pass the test.
Forecasting Future World Events with Neural Networks
We test language models on our forecasting task and find that performance is far below a human expert baseline.
Beyond the Imitation Game: Quantifying and extrapolating the capabilities of language models
BIG-bench focuses on tasks that are believed to be beyond the capabilities of current language models.
Teaching Models to Express Their Uncertainty in Words
We show that a GPT-3 model can learn to express uncertainty about its own answers in natural language -- without use of model logits.
Truthful AI: Developing and governing AI that does not lie
Establishing norms or laws of AI truthfulness will require significant work to: (1) identify clear truthfulness standards; (2) create institutions that can judge adherence to those standards; and (3) develop AI systems that are robustly truthful.
TruthfulQA: Measuring How Models Mimic Human Falsehoods
We crafted questions that some humans would answer falsely due to a false belief or misconception.
Ranked #15 on
Question Answering
on TruthfulQA
Active Reinforcement Learning: Observing Rewards at a Cost
Active reinforcement learning (ARL) is a variant on reinforcement learning where the agent does not observe the reward unless it chooses to pay a query cost c > 0.
Sensory Optimization: Neural Networks as a Model for Understanding and Creating Art
How is it that humans are able to understand and create stylized representations?
Cultural Vocal Bursts Intensity Prediction
Object Recognition
+1
Generalizing from a few environments in safety-critical reinforcement learning
Before deploying autonomous agents in the real world, we need to be confident they will perform safely in novel situations.
Active Reinforcement Learning with Monte-Carlo Tree Search
Active Reinforcement Learning (ARL) is a twist on RL where the agent observes reward information only if it pays a cost.
The Malicious Use of Artificial Intelligence: Forecasting, Prevention, and Mitigation
This report surveys the landscape of potential security threats from malicious uses of AI, and proposes ways to better forecast, prevent, and mitigate these threats.
Trial without Error: Towards Safe Reinforcement Learning via Human Intervention
We formalize human intervention for RL and show how to reduce the human labor required by training a supervised learner to imitate the human's intervention decisions.
When Will AI Exceed Human Performance? Evidence from AI Experts
Advances in artificial intelligence (AI) will transform modern life by reshaping transportation, health, science, finance, and the military.
Agent-Agnostic Human-in-the-Loop Reinforcement Learning
Providing Reinforcement Learning agents with expert advice can dramatically improve various aspects of learning.
Learning the Preferences of Ignorant, Inconsistent Agents
If we assume that choices are approximately optimal according to some utility function, we can treat preference inference as Bayesian inverse planning.
Help or Hinder: Bayesian Models of Social Goal Inference
Everyday social interactions are heavily influenced by our snap judgments about others goals.
