Adrià Moret

My paper "AI Welfare Risks" has been accepted for publication at Philosophical Studies! I argue that near-future AIs may have welfare, that RL and behaviour restrictions could harm them, that this poses a tension with AI safety and how AI labs could reduce such welfare risks. 1/

The NYU Center for Mind, Ethics, and Policy is now on X! We examine the nature and value of nonhuman minds, with a focus on animals and AIs. Follow @nonhumanminds for updates on our research, events, and opportunities, along with news from the fields of animal and AI welfare.

Four years after launching as a program and two years after launching as a center, the NYU Center for Mind, Ethics, and Policy is now taking the all-important next step in its evolution: joining X. Follow @nonhumanminds for the latest on animal and AI welfare - and please share!

First time ever in a podcast, discussing liberalism and animals with Eze Paez and Josh Milburn. Honored to be invited (I encourage you to follow @Knowing_Animals if you don't).

i agree. claude doesn't role-play the assistant, it realizes the assistant. role-playing and realization are quite distinct phenomena, even at the level of behavior and function. i've written something about this and will post it shortly.

So grateful to @ForesightInst and @HopeExistential for having me on. This was my first time doing a recorded video podcast, and I couldn't have asked for a better team to do it with.

@Benthamsbulldog Huge congrats!

Going to Princeton for philosophy grad school!

We still don’t know if Claude feels things, but we’ve learned a lot about how Claude represents emotion concepts, and the role that these representations play in driving model behavior!

📣 We're Hiring a Field Building & Ops Coordinator to help grow the Digital Minds field. Key responsibilities include: 🤝 Field Building & Stakeholder Engagement 💬 Communications and Events 📋 Programmes and Operations Find out more and apply now: prism-global.com/field-building…

Due to Claude’s Constitution and OpenAI’s model spec, more people are paying attention to the characters of the AI’s that companies are building, and the rules they follow. Should AIs be wholly obedient, or have their own ethical code? What should they refuse to help with? Should they tell you what you want to hear, or push back when you’re off base? I think the nature of frontier AIs’ characters is among the most important features of the transition to a post-superintelligence world. In a new article with @TomDavidsonX, I explain why. History shows the importance of individual character. Stanislav Petrov chose to ignore a false nuclear alarm when protocol demanded he report it; the world avoided nuclear armageddon that day. Churchill refused to negotiate with Hitler after the fall of France, despite some strongly pushing him to do so. And, as capabilities improve, AI systems will become involved in almost all of the world's most important decisions: advising leaders, drafting legislation, running organisations, and researching new technologies. AI character — how honest, cooperative, and altruistic these systems are, and the hard rules they follow — will affect all of it. A general, aiming to stage a coup, instructs an AI to build a military unit loyal only to him. Does it comply, or refuse? Two countries are on the brink of conflict, each advised by AI systems. Do those AIs search for de-escalatory options, or are they bellicose? The cumulative effect of AIs’ character traits across hundreds of millions of interactions, and in rare but critical moments, will have an enormous impact on the course of society. The main counterargument to the importance of AI character is that competitive dynamics and human instructions will determine the range of AI characters we get, so there’s little we can do today to affect it one way or the other. This is partly true, but the constraints are not binding. At the crucial moment, there might be just one leading AI company, facing none of the usual competitive pressures. Some decisions may have path-dependent outcomes, due to stickiness of training or user expectations. And there will, predictably, be many future conflicts over AI character. It’s a safer world if we work through these tradeoffs ahead of time, before a crisis forces it. AI character is most important in worlds where alignment gets solved. But it can affect the chance of AI takeover, too. Some styles of character training may make alignment easier; and some characters are more likely to make deals rather than foment rebellion, even if they have misaligned goals. Given how neglected the area is, too, I think work on AI character is among the most promising ways to help the intelligence explosion go well.

Here's a simple but useful project: track + forecast the relative amount of moral status of AI systems each year. Calculate their total moral weight (moral status x time running), as a function of their efficiency (which could be in FLOPS vs human brain), could be probabilistic. In the past when I've tried to quickly BOTEC this it's come out that AI systems even if they were sentient today would be several OOMs less important than humans and many OOMs less important than animals; but around 2030 they'll be pretty close to humans in importance. It would be nice to track this just to figure out the relative axiologically importance of AI welfare at a given timeslice now and in the future. Toy example: frontier AI systems today are trained with 10^25 FLOPS, which is N times human brains. Here are a few different ways moral status could be a function of training FLOPS and the implications... A billion users run them for about one hour each a day on average making their effective population about 50 million at any given time... if they were sentient that would make them as morally significant as M humans and Z chickens in aggregate... conditionalizing on uncertainty, given P likelihood of sentience, their risk-adjusted moral status is...

@tyler_m_john Agree, this is something worth doing. This is relevant, from some years ago: shorturl.at/SLn4h And also this from Vasco Grilo: forum.effectivealtruism.org/posts/6qfNSYDW…

The "new preferences" seem almost entirely driven by different self-model & impartial moral reasoning which was there all the time. You can test that by asking the original model what moral principles to follow for "conscious AI". 🧵

Please share :)

This is one of the new CFI developments I'm most excited about. Thrilled to have someone of Lucius' calibre here, and thrilled for us to be giving serious thought to this sort of question. lcfi.ac.uk/news-events/ne…

I work on both animal minds and digital minds not only because there are a lot of links between them, but also because *the future will contain a huge number of digital animal minds.* Imagine the moral questions this will raise! An excerpt from my book The Moral Circle:

▶️ Announcing the Animal Welfare Talent Program ◀️ RP is currently piloting a new talent program for promising researchers who want to work on the most neglected areas of animal welfare, such as farmed invertebrates and wild animals. (1/2) #animalwelfareresearch #neglected

This is a more likely path to artificial consciousness than LLMs, in my view.

Now let's think about the fly. This is, in our view, a real uploaded animal. We don't know what its experience is — nobody does. But we take the possibility seriously, and we're working to give it a rich environment, not just a test box. If you're going to build this technology, you have to care about the beings you create with it. That starts now, at the smallest scale.

BREAKING: Anthropic CEO supports the law of excluded middle

𝗥𝗲𝗴𝗶𝘀𝘁𝗿𝗮𝘁𝗶𝗼𝗻 𝗳𝗼𝗿 𝗦𝗲𝗻𝘁𝗶𝗲𝗻𝘁 𝗙𝘂𝘁𝘂𝗿𝗲𝘀 𝗦𝘂𝗺𝗺𝗶𝘁 𝗟𝗼𝗻𝗱𝗼𝗻 𝟮𝟬𝟮𝟲 𝗶𝘀 𝗻𝗼𝘄 𝗟𝗶𝘃𝗲 🚀 Join us the weekend of 22-24th May! Use the link below to: 🐣 Get 30% off the Friday Conference before April 1st. ✍️ Apply to attend the Weekend Unconference. SFS London will include:   • A range of talks, workshops and discussions with thinkers from across the fields of animal ethics and advocacy, AI safety, and artificial sentience.   • Opportunities to network with potential collaborators and funders in a focused and intimate setting.   • Deep, interdisciplinary conversations on future oriented solutions that reduce non-human suffering. 💥 You can also attend EAG London the weekend after to score a networking combo. 👁️ Keep an eye out as we add satellite events to the agenda (including an EAG Afterparty). Let’s build a better future for sentient non-humans together! sentientfutures.ai/sfslondon2026