Value Alignment

We can learn human values from data the same way we learn everything else

Greg Brockman

“Value alignment may be tractable because we have already built systems that learn things humans cannot explicitly specify - like recognizing cats. Learning human preferences from data and feedback follows the same principle, and early results at OpenAI were encouraging.”

We don't even know what we want for ourselves

Meia Chita-Tegmark

“Value alignment is doubly hard because humans often lack access to their own preferences - competing desires, subconscious motivations, and impulse control all complicate what it means to learn what humans really want.”

King Midas and the problem of getting what you ask for

Meia Chita-Tegmark

“Value alignment is complicated because people disagree about values, cultures differ, and - as King Midas learned - we need to be careful what we ask for. The problem is even deeper because often we don't know what we really want for ourselves, let alone how to ask an AI to help us get it.”

Iason Gabriel on Foundational Philosophical Questions in AI Alignment

Iason Gabriel

“might think about aligning ai systems with human values and i wrote this paper because i wanted to bring different communities together so on the one hand i wanted to show machine learning researchers that there were some interesting normative questions questi”

Eliezer Yudkowsky – AI Alignment: Why It's Hard, and Where to Start

Eliezer Yudkowsky

“…on was like as a neuroscientist can I look for analogues of value alignment problems and in my own work and if so how that's a new question if I'm not allowed to take five minutes to go quiet and think about it then my immediate answer is like it's not obvious”

AI, Ethics, and the Value Alignment Problem with Meia Chita-Tegmark and Lucas Perry

Meia Chita-Tegmark

“…p a lot when we talk about AI and ethics is this concept of value alignment and so I was hoping you could both maybe talk just a minute about what value alignment is and why it's important to this question of AI and ethics so a value alignment in my view is br”

AI, Ethics, and the Value Alignment Problem with Meia Chita-Tegmark and Lucas Perry

Meia Chita-Tegmark

“like the next ten years whereas value alignments really thinking about the instantiation of ethics and machines and making machine systems that are corrigible and robust and docile which will create a world that we're all happy about living in okay so I think”

Iason Gabriel on Foundational Philosophical Questions in AI Alignment

Iason Gabriel

“…re we have a lot of public thought and discussion about the values that would make a system like that fair and the challenge then is to work out whether there is a technical approximation of these values that satisfactory realizes them in a way that conduces t”

AI and Value Alignment | Jaan Tallinn

Jaan Tallinn

“…d a way to figure out what Humanity wants when I talk about value alignment on in public one question I almost always get is whose values are we talking about and in response I usually kind of Mumble something about the vast majority of human values being so o”

AI and Value Alignment | Jaan Tallinn

Jaan Tallinn

“…st over half a century since the original warnings about AI value alignment by Alan touring and Norbert Wier I certainly hope that we have still another 50 but I know that several experts in this room are much more optimistic about AI timelines and thus pessim”

Evan Hubinger on Inner Alignment, Outer Alignment, and Proposals for Building Safe Advanced AI

Evan Hubinger

“…tuff you don't want it to do and so outer alignment is this value alignment problem of how do you find objectives which are actually good to optimize to optimize to optimize but then even if you have found an objective which is actually good to optimize optimi”