226 citations · 690 across the 20 of their papers we have counts for
20 papers
Debate Training Reduces Reward Hacking in RLAIF
Zachary Kenton, Lili Janzer, Rory Greig +8
We demonstrate that RL finetuning an LLM using debate, a two-player adversarial game between a generator and a critic adjudicated by a weaker LLM judge, reduces reward hacking comp…
An Approach to Technical AGI Safety and Security
Rohin Shah, Alex Irpan, Alexander Matt Turner +27
Artificial General Intelligence (AGI) promises transformative benefits but also presents significant risks. We develop an approach to address the risk of harms consequential enough…
On scalable oversight with weak LLMs judging strong LLMs
Zachary Kenton, Noah Y. Siegel, János Kramár +8
Scalable oversight protocols aim to enable humans to accurately supervise superhuman AI. In this paper we study debate, where two AI's compete to convince a judge; consultancy, whe…
The Ethics of Advanced AI Assistants
Iason Gabriel, Arianna Manzini, Geoff Keeling +54
This paper focuses on the opportunities and the ethical and societal risks posed by advanced AI assistants. We define advanced AI assistants as artificial agents with natural langu…
A Mechanism-Based Approach to Mitigating Harms from Persuasive Generative AI
Seliem El-Sayed, Canfer Akbulut, Amanda McCroskery +17
Recent generative AI systems have demonstrated more advanced persuasive capabilities and are increasingly permeating areas of life where they can influence decision-making. Generat…
Challenges with unsupervised LLM knowledge discovery
Sebastian Farquhar, Vikrant Varma, Zachary Kenton +3
We show that existing unsupervised methods on large language model (LLM) activations do not discover knowledge -- instead they seem to discover whatever feature of the activations…
Explaining grokking through circuit efficiency
Vikrant Varma, Rohin Shah, Zachary Kenton +2
One of the most surprising puzzles in neural network generalisation is grokking: a network with perfect training accuracy but poor generalisation will, upon further training, trans…
Goal Misgeneralization: Why Correct Specifications Aren't Enough For Correct Goals
Rohin Shah, Vikrant Varma, Ramana Kumar +4
The field of AI alignment is concerned with AI systems that pursue unintended goals. One commonly studied mechanism by which an unintended goal might arise is specification gaming,…
Discovering Agents
Zachary Kenton, Ramana Kumar, Sebastian Farquhar +3
Causal models of agents have been used to analyse the safety aspects of machine learning systems. But identifying agents is non-trivial -- often the causal model is just assumed by…
Safe Deep RL in 3D Environments using Human Feedback
Matthew Rahtz, Vikrant Varma, Ramana Kumar +3
Agents should avoid unsafe behaviour during both training and deployment. This typically requires a simulator and a procedural specification of unsafe behaviour. Unfortunately, a s…
Ethical and social risks of harm from Language Models
Laura Weidinger, John Mellor, Maribeth Rauh +20
This paper aims to help structure the risk landscape associated with large-scale Language Models (LMs). In order to foster advances in responsible innovation, an in-depth understan…
Alignment of Language Agents
Zachary Kenton, Tom Everitt, Laura Weidinger +3
For artificial intelligence to be beneficial to humans the behaviour of AI agents needs to be aligned with what humans want. In this paper we discuss some behavioural issues for la…
Imitating Interactive Intelligence
Josh Abramson, Arun Ahuja, Iain Barr +26
A common vision from science fiction is that robots will one day inhabit our physical spaces, sense the world as we do, assist our physical labours, and communicate with us through…
A Systematic Comparison of Bayesian Deep Learning Robustness in Diabetic Retinopathy Tasks
Angelos Filos, Sebastian Farquhar, Aidan N. Gomez +6
Evaluation of Bayesian deep learning (BDL) methods is challenging. We often seek to evaluate the methods' robustness and scalability, assessing whether new tools give `better' unce…
Generalizing from a few environments in safety-critical reinforcement learning
Zachary Kenton, Angelos Filos, Owain Evans +1
Before deploying autonomous agents in the real world, we need to be confident they will perform safely in novel situations. Ideally, we would expose agents to a very wide range of…
On the Relation Between the Sharpest Directions of DNN Loss and the SGD Step Length
Stanisław Jastrzębski, Zachary Kenton, Nicolas Ballas +3
Stochastic Gradient Descent (SGD) based training of neural networks with a large learning rate or a small batch-size typically ends in well-generalizing, flat regions of the weight…
Three Factors Influencing Minima in SGD
Stanisław Jastrzębski, Zachary Kenton, Devansh Arpit +4
We investigate the dynamical and convergent properties of stochastic gradient descent (SGD) applied to Deep Neural Networks (DNNs). Characterizing the relation between learning rat…
The squeezed limit of the bispectrum in multi-field inflation
Zachary Kenton, David J. Mulryne
We calculate the squeezed limit of the bispectrum produced by inflation with multiple light fields. To achieve this we allow for different horizon exit times for each mode and calc…
Generating the cosmic microwave background power asymmetry with
Zachary Kenton, David J. Mulryne, Steven Thomas
We consider a higher order term in the expansion for the CMB power asymmetry generated by a superhorizon isocurvature field fluctuation. The term can generate the asymmetry wi…
D-brane Potentials in the Warped Resolved Conifold and Natural Inflation
Zachary Kenton, Steven Thomas
In this paper we obtain a model of Natural Inflation from string theory with a Planckian decay constant. We investigate D-brane dynamics in the background of the warped resolved co…