AI Safety Fundamentals: Alignment 201

Alternative title: “When should you assume that what could go wrong, will go wrong?” Thanks to Mary Phuong and Ryan Greenblatt for helpful suggestions and discussion, and Akash Wasil for some edits. In discussions of AI safety, people often propose the assumption that something goes as badly as possible. Eliezer Yudkowsky in particular has argued for the importance of security

Previously, I argued that emergent phenomena in machine learning mean that we can’t rely on current trends to predict what the future of ML will be like. In this post, I will argue that despite this, empirical findings often do generalize very far, including across “phase transitions” caused by emergent behavior.This might seem like a contradiction, but actually I think diver

Right now I’m working on finding a good objective to optimize with ML, rather than trying to make sure our models are robustly optimizing that objective. (This is roughly “outer alignment.”) That’s pretty vague, and it’s not obvious whether “find a good objective” is a meaningful goal rather than being inherently confused or sweeping key distinctions under the rug. So

Using hard multiple-choice reading comprehension questions as a testbed, we assess whether presenting humans with arguments for two competing answer options, where one is correct and the other is incorrect, allows human judges to perform more accurately, even when one of the arguments is unreliable and deceptive. If this is helpful, we may be able to increase our justified trust in

Chain-of-thought prompting has demonstrated remarkable performance on various natural language reasoning tasks. However, it tends to perform poorly on tasks which requires solving problems harder than the exemplars shown in the prompts. To overcome this challenge of easy-to-hard generalization, we propose a novel prompting strategy, least-to-most prompting. The key idea in this str

This paper presents a technique to scan neural network based AI models to determine if they are trojaned. Pre-trained AI models may contain back-doors that are injected through training or by transforming inner neuron weights. These trojaned models operate normally when regular inputs are provided, and mis-classify to a specific output label when the input is stamped with some spec

This post tries to explain a simplified version of Paul Christiano’s mechanism introduced here, (referred to there as ‘Learning the Prior’) and explain why a mechanism like this potentially addresses some of the safety problems with naïve approaches. First we’ll go through a simple example in a familiar domain, then explain the problems with the example. Then I’ll discus

It would be very convenient if the individual neurons of artificial neural networks corresponded to cleanly interpretable features of the input. For example, in an “ideal” ImageNet classifier, each neuron would fire only in the presence of a specific visual feature, such as the color red, a left-facing curve, or a dog snout. Empirically, in models we have studied, some of the n

Abstract: Existing techniques for training language models can be misaligned with the truth: if we train models with imitation learning, they may reproduce errors that humans make; if we train them to generate text that humans rate highly, they may output errors that human evaluators can't detect. We propose circumventing this issue by directly finding latent knowledge inside

The field of reinforcement learning (RL) is facing increasingly challenging domains with combinatorial complexity. For an RL agent to address these challenges, it is essential that it can plan effectively. Prior work has typically utilized an explicit model of the environment, combined with a specific planning algorithm (such as tree search). More recently, a new family of methods