Is this you? As a journalist, you can create a free Muck Rack account to customize your profile, list your contact preferences, and upload a portfolio of your best work.
Claim your profile
Articles by Joar Skalse
How to Contribute to Theoretical Reward Learning Research — LessWrong
This is the eighth (and, for now, final) post in the theoretical reward learning sequence, which starts in this post. Here, I will provide a few pointers to anyone who might be interested in contributing to further work on this research agenda, in the form of a few concrete and shovel-ready open problems, a few ideas on how those problems may be approached, and a few useful general insights about this problem setting.
Other Papers About the Theory of Reward Learning — LessWrong
This is the seventh post in the theoretical reward learning sequence, which starts in this post. Here, I will provide shorter summaries of a few additional papers on the theory of reward learning, but without going into as much depth as I did in the previous posts (but if there is sufficient demand, I might extend these summaries into their own posts). The original motivation for this paper came from a desire to spite a guy I met at IJCAI in 2022.
Partial Identifiability in Reward Learning — LessWrong
In this post, I will provide a summary of the paper Invariance in Policy Optimisation and Partial Identifiability in Reward Learning, and explain some of its results. I will assume basic familiarity with reinforcement learning. This is the second post in the theoretical reward learning sequence, which starts in this post (though this post is self-contained). This was joint work together with Matthew Farrugia-Roberts and Adam Gleave.
STARC: A General Framework For Quantifying Differences Between Reward Functions — LessWrong
In this post, I will provide a summary of the paper STARC: A General Framework For Quantifying Differences Between Reward Functions, and explain some of its results. I will assume basic familiarity with reinforcement learning. This is the fourth post in the theoretical reward learning sequence, which starts in this post (though this post is self-contained). In this paper, we consider the question of how to quantify the distance between reward functions in an informative way.
Defining and Characterising Reward Hacking — LessWrong
In this post, I will provide a summary of the paper Defining and Characterising Reward Hacking, and explain some of its results. I will assume basic familiarity with reinforcement learning. This is the sixth post in the theoretical reward learning sequence, which starts in this post (though this post is self-contained). This was joint work together with Nikolaus Howe, Dmitrii Krasheninnikov, and David Krueger.
Misspecification in Inverse Reinforcement Learning - Part II — LessWrong
In this post, I will provide a summary of the paper Quantifying the Sensitivity of Inverse Reinforcement Learning to Misspecification, and explain some of its results. I will assume basic familiarity with reinforcement learning. This is the fifth post in the theoretical reward learning sequence, which starts in this post. This post is somewhat self-contained, but I will largely assume that you have read this post and this post before reading this one.
Misspecification in Inverse Reinforcement Learning — LessWrong
In this post, I will provide a summary of the paper Misspecification in Inverse Reinforcement Learning, and explain some of its results. I will assume basic familiarity with reinforcement learning. This is the third post in the theoretical reward learning sequence, which starts in this post (though this post is self-contained). This paper won the best paper award at AAAI 2023.
The Theoretical Reward Learning Research Agenda: Introduction and Motivation — LessWrong
At the time of writing, I have just (nearly) finished my PhD at Oxford. During that time, most of my main research has been motivated by the goal of developing a theoretical foundation for the field of reward learning. The purpose of this sequence is to explain and motivate this research agenda, and to provide an accessible overview over some of my results.
Towards Guaranteed Safe AI: A Framework for Ensuring Robust and Reliable AI Systems — LessWrong
I want to draw attention to a new paper, written by myself, David "davidad" Dalrymple, Yoshua Bengio, Stuart Russell, Max Tegmark, Sanjit Seshia, Steve Omohundro, Christian Szegedy, Ben Goldhaber, Nora Ammann, Alessandro Abate, Joe Halpern, Clark Barrett, Ding Zhao, Tan Zhi-Xuan, Jeannette Wing, and Joshua Tenenbaum. In this paper we introduce the concept of "guaranteed safe (GS) AI", which is a broad research strategy for obtaining safe AI systems with provable quantitative safety guarantees.
#ICLR2023 invited talk: Data, history and equality with Elaine Nsoesie - ΑΙhub
Figure from Use of Deep Learning to Examine the Association of the Built Environment With Prevalence of Neighborhood Adult Obesity, Adyasha Maharana and Elaine Okanyene Nsoesie. Image on the right represents actual obesity prevalence; on the left, cross-validated estimates of obesity prevalence based on features of the built environment extracted from satellite images. Figure reproduced under CC-BY licence.
Forthcoming machine learning and AI seminars: May 2023 edition - ΑΙhub
This post contains a list of the AI-related seminars that are scheduled to take place between 10 May and 30 June 2023. All events detailed here are free and open for anyone to attend virtually. 10 May 2023 Natural Language Generation Problems and Challenges Speaker: Konstantinos Diamantaras Organised by: Chalmers AI Research Centre Zoom link is here. AI in programming education Speaker: Simon McCallum and Arne Styve Organised by: Norwegian Open AI Lab Register here.
Machine learning helps researchers separate compostable from conventional plastic waste - ΑΙhub
By Deborah Pirchner Disposable plastics are everywhere: Food containers, coffee cups, plastic bags. Some of these plastics, called compostable plastics, can be engineered to biodegrade under controlled conditions. However, they often look identical to conventional plastics, get recycled incorrectly and, as a result, contaminate plastic waste streams and reduce recycling efficiency. Similarly, recyclable plastics are often mistaken for compostable ones, resulting in polluted compost.
The basic reasons I expect AGI ruin
Info about AGI propagates too slowly through the field, because when one ML person updates, they usually don't loudly share their update with all their peers. This is because: 1. AGI sounds weird, and they don't want to sound like a weird outsider. 2. Their peers and the community as a whole might perceive this information as an attack on the field, an attempt to lower its status, etc. 3.
[1906.01820] Risks from Learned Optimization in Advanced Machine Learning Systems
[Submitted on 5 Jun 2019 (v1), last revised 1 Dec 2021 (this version, v3)] Authors:Evan Hubinger, Chris van Merwijk, Vladimir Mikulik, Joar Skalse, Scott Garrabrant Download PDF Abstract: We analyze the type of learned optimization that occurs when a learned model (such as a neural network) is itself an optimizer - a situation we refer to as mesa-optimization, a neologism we introduce in this paper.
Risks from Learned Optimization: Introduction
Crossposted from the AI Alignment Forum. May contain more technical jargon than usual. This is the first of five posts in the Mesa-Optimization Sequence based on the upcoming MIRI paper “Risks from Learned Optimization in Advanced Machine Learning Systems” by Evan Hubinger, Chris van Merwijk, Vladimir Mikulik, Joar Skalse, and Scott Garrabrant.
Show More
loading
Actions
Is this you?
As a journalist, you can create a free Muck Rack account to customize your profile, list your contact preferences, and upload a portfolio of your best work.Get in touch with Joar
Contact Joar, search articles and posts on X, monitor coverage, and track replies from one place.
Learn more about Muck Rack