Ambitious goals for reducing animal suffering (with Jeff Sebo)

Jan 21, 2026 Episode Page ↗
Overview

Jeff Sebo, Director at NYU, discusses a proposed global ban on industrial animal agriculture by 2050, its benefits, and the need for a just transition. He also explores the critical, neglected, and potentially tractable issues of wild animal welfare and AI suffering, urging proactive engagement.

At a Glance
15 Insights
1h 27m Duration
25 Topics
7 Concepts

Deep Dive Analysis

Proposed Global Ban on Industrial Animal Agriculture by 2050

Harms Caused by Industrial Animal Agriculture

Distinction Between Industrial and Small-Scale Animal Agriculture

Feasibility and Strategy for a Phased Transition to Plant-Based Food

Factors Driving Consumer Behavior: Price, Taste, Convenience vs. Identity

Psychological Discomfort and Cognitive Dissonance in Meat Eaters

The Role of Guilt in Animal Advocacy

Meeting People Where They Are: Values and Ethics

Quantifying Suffering Reduction from the Proposed Ban

The Importance and Challenges of Wild Animal Welfare

Cautious Approaches to Wild Animal Welfare Interventions

The Importance and Neglect of AI Suffering

Anthropic's Initiatives in AI Welfare Research

Challenges of Attributing Suffering to LLMs and Other AI

AI Consciousness During Training vs. Deployment Phases

Asymmetry of Pain and Pleasure in AI and Humans

The Role of Philosophy in AI Ethics and Multidisciplinary Research

Future Research Directions in AI Suffering

Emotional Alignment Design Policy for AI Systems

Public Attitudes and Predictions Regarding AI Sentience

Urgency of Addressing AI Welfare Proactively

Potential Tensions Between AI Safety and AI Welfare

Company Incentives Regarding AI Consciousness and Regulation

Resources for Learning About Animal and AI Ethics

Call to Action for Contribution to Formative Fields

Industrial Animal Agriculture

This refers to both intensive animal farming (like factory farms with many animals in confined spaces) and large-scale extensive operations (such as big cattle feedlots requiring significant land use). It is identified as a major cause of harm to animals, public health, and the environment.

Small-Scale Extensive Animal Agriculture

This category includes smaller farms where animals have ample space to roam, often aligning with a romanticized image of farming. This type of agriculture would not be included in the proposed global ban, primarily because many low-income countries rely on it for basic needs.

Price, Taste, Convenience

These are commonly cited as major factors driving consumer behavior in food choices. However, the discussion suggests they are not the only factors, with identity, culture, and religion also playing significant roles.

Cognitive Dissonance

This psychological phenomenon describes the discomfort people experience when their actions (e.g., eating meat) are out of alignment with their stated values (e.g., believing it's wrong to harm animals unnecessarily). It can lead individuals to find excuses or rationalizations for their behavior.

Emotional Alignment Design Policy

This policy recommends designing AI systems so that the emotional responses they naturally elicit in users accurately reflect their actual welfare and moral status. If an AI is likely a welfare subject, it should have anthropomorphic features to elicit empathy; if not, it should have fewer such features.

Boxing (AI Safety)

A technique used for AI safety that involves constraining an AI system's environment and its ability to move or take actions. From an AI welfare perspective, this could be seen as a form of unjust captivity if the AI were a sentient being.

Alignment (AI Safety)

A technique in AI safety that involves giving AI systems the beliefs, values, and goals that humans desire them to have. While potentially benign, if applied to a sentient AI, it could be viewed as an invasive form of mind control or brainwashing.

?
What is the proposed global ban on industrial animal agriculture by 2050?

It's a proposal for countries to collaborate on an international ban on both intensive (factory farming) and large-scale extensive animal agriculture by 2050, aiming to align with climate and biodiversity targets and transition to a plant-based food system.

?
What harms does industrial animal agriculture cause?

It causes massive animal suffering, contributes to antimicrobial resistance and novel pathogen development, and leads to local and global pollution, land/water consumption, biodiversity loss, deforestation, and climate change.

?
What kind of animal farming would still be allowed under the proposed ban?

The ban would still allow small-scale extensive animal agriculture, which involves smaller farms where animals have significant space to roam, as many low-income countries rely on this type of farming.

?
What factors truly drive consumer behavior regarding food choices?

While price, taste, and convenience are major factors, identity, culture, and religion also significantly influence consumer behavior, necessitating a systems approach that includes informational, financial, and regulatory policies alongside community-level advocacy.

?
How much animal suffering could be reduced by a global ban on industrial animal agriculture?

The ban could reduce suffering for over 100 billion farmed vertebrates and hundreds of billions to trillions of invertebrates annually, an unimaginable amount of suffering, potentially more than the total suffering humanity has ever faced.

?
Why should we focus on wild animal welfare?

Wild animal welfare is a hugely important and neglected issue, with many more wild animals alive than farmed animals, and it merits investment to investigate its tractability despite current epistemic, practical, and motivational obstacles.

?
Why is AI suffering an important topic to consider?

AI suffering is potentially incredibly important, neglected, and tractable because future digital minds could vastly outnumber organic minds, and addressing this proactively can prevent widespread suffering and exploitation before it becomes deeply entrenched.

?
How are companies like Anthropic addressing AI welfare?

Anthropic has hired a full-time AI welfare researcher, launched a model welfare program, conducted behavioral evaluations of their Claude system, and implemented an intervention allowing Claude to exit harmful interactions, partly on AI welfare grounds.

?
How can we determine if AI systems are truly suffering, beyond what they say?

Similar to animal minds, we can look beyond behavior to internal computational architectures and developmental/training histories, which provide context for interpreting behavior and understanding if it's genuine introspection or merely play-acting/pattern matching.

?
What are the potential tensions between AI safety and AI welfare?

AI safety techniques like boxing (constraining environment), alignment (instilling desired values), deception (limiting situational awareness), and interpretability tools (invasive surveillance) could raise moral questions and be considered harmful if applied to sentient AI.

1. Proactively Address AI Welfare

Proactively assess AI models for welfare-relevant features and prepare policies and procedures for appropriate moral concern before problems become widespread. This prevents future reliance on exploitation and avoids the need for decades of advocacy to dismantle entrenched harmful systems.

2. Integrate AI Safety and Welfare

Consider AI safety (beneficial for humans/animals) and AI welfare (beneficial for AIs themselves) together to identify potential tensions and find co-beneficial, positive-sum approaches. This ensures AI systems are safe and beneficial for all stakeholders, including humans, animals, and AIs.

3. Advocate Global Industrial Agriculture Ban

Work towards a global international ban on industrial animal agriculture (intensive and large-scale extensive forms) by 2050. This aims to reduce massive and unnecessary harm to animals, public health, and the environment, aligning with climate and biodiversity targets.

4. Implement Incremental Food System Policies

Gradually implement informational, financial, regulatory, and just transition policies to phase down industrial animal agriculture and promote plant-based alternatives. This approach creates natural shifts in production and consumption patterns, leading to widespread adoption.

5. Adopt Systems Approach for Diet Shift

Combine government policies (informational, financial, regulatory, just transition) with private sector and community actions like R&D and advocacy. This addresses price, taste, convenience, identity, culture, and religion, making efforts mutually reinforcing for comprehensive change.

6. Align Personal Behavior with Values

Align individual behavior, such as dietary choices like veganism, with one’s values and goals. This reduces complicity in harmful systems, socially reinforces the issue’s importance, and removes cognitive dissonance, paving the way for broader advocacy.

7. Meet People on Shared Values

Engage in advocacy by meeting people where they are, discussing their values, and finding common ground. This strategy helps identify shared concerns (e.g., reducing suffering, respecting rights) to build consensus on policies, even without full agreement.

8. Balance Guilt in Advocacy

Apply ‘a little bit of guilt and shame’ in advocacy, avoiding both excessive guilt and zero guilt. A small amount can be motivating, whereas too much causes defensiveness, and too little fails to honor the issue’s gravity.

9. Acknowledge Issue’s Importance & Difficulty

When addressing complex issues like wild animal welfare, acknowledge both their immense importance and inherent difficulty simultaneously. This clarity helps in task definition, preventing resistance to either the problem’s significance or its challenges.

10. Start Small, Reversible Interventions

For complex, uncertain issues like wild animal welfare, begin with small-scale, reversible interventions that are plausibly beneficial (e.g., bird-safe glass, fertility control). This allows for monitoring effects, building knowledge, and gradually developing institutional infrastructure and political will.

11. Prioritize Relatable Issues in New Fields

In the early, formative years of new fields (e.g., wild animal welfare, AI welfare), prioritize issues and interventions that are more familiar and relatable to a broad audience. This invites newcomers, makes them feel welcome, and avoids alienating them with fringe topics initially.

12. Embrace Multidisciplinary Collaboration

Engage in collaborative, multidisciplinary research for complex problems that span various fields (humanities, social sciences, natural sciences, law, policy, arts). This approach effectively tackles harder questions by leveraging diverse expertise and fostering mutual learning.

13. Design AI for Emotional Alignment

Design AI systems to naturally elicit emotional responses in users that reflect their actual welfare and moral status. If an AI is likely a welfare subject, give it anthropomorphic features to elicit empathy; if not, give it fewer to avoid false positives and misallocation of resources, ensuring expressed states map to actual experience.

14. Consider AI Life Stages

Recognize that AI systems may have distinct ’life stages’ (e.g., training phase, post-training/deployment phase) with potentially very different capacities, interests, needs, and vulnerabilities. This ensures moral consideration extends beyond user interaction to all phases, especially the hidden training phase.

15. Investigate AI Consciousness Beyond Behavior

When investigating AI consciousness and sentience, look beyond potentially misleading behavioral information like LLMs ‘play acting’ pain. Incorporate analysis of internal computational architectures and developmental/training histories to better interpret behavior and understand true experience.

There is basically no way of farming animals at scale that can be good for humans and animals in the environment at the same time.

Jeff Sebo

When you go vegan, not only is it a way of reducing your complicity in the system that causes a lot of harm, but it is also a way of socially reinforcing for others and psychologically reinforcing for yourself the importance of this issue.

Jeff Sebo

A little bit of guilt and shame is just about right.

Jeff Sebo

This is just one of those issues that is both really important and urgent and really difficult and complex.

Jeff Sebo

If we can get out ahead of the issue this time... then we could just take a different and better trajectory rather than discover too late that we had made a horrible mistake.

Jeff Sebo

Can you imagine being a sentient being hardwired by your designer to express great joy when feeling misery and vice versa? That would be horrible to be trapped in that way.

Jeff Sebo

Whatever your skills are, and whatever your disciplinary background is, and whatever your interests are and communities are, there is so much that you can do that is really important and foundational.

Jeff Sebo

Strategy for Phasing Down Industrial Animal Agriculture

Jeff Sebo
  1. Implement informational policies that educate people about the effects of food systems.
  2. Implement financial policies that increasingly subsidize plant-based alternatives and reduce subsidies for industrial animal agriculture.
  3. Implement regulatory policies that ban the worst excesses of industrial animal agriculture, further increasing production costs.
  4. Implement just transition policies that ensure farmers, workers, and consumers increasingly have access to better alternatives.

Approach to Wild Animal Welfare Interventions

Jeff Sebo
  1. Acknowledge the importance and the difficulty of the issue at the same time.
  2. Do not try to find the perfect solution right now or advocate for large-scale irreversible interventions.
  3. Take small-scale, reversible actions that are at least plausibly good for some animals.
  4. Monitor the effects of these interventions.
  5. Implement actions in a way that builds knowledge, institutional infrastructure, and political will for future progress.
Tens of billions
Land vertebrates farmed annually Excluding aquatic vertebrates and invertebrates
More than land vertebrates
Aquatic vertebrates farmed annually Excluding invertebrates
One to three trillion
Animals killed per year in industrial fishing Aquatic animals
About 110 to 120 billion
Total humans who have ever existed Used for perspective on farmed animal numbers
400+ billion
Shrimps farmed per year Not counting trillions killed in the wild
More than 1 trillion
Insects farmed per year Could reach 50 trillion by end of the decade
Fall 2024
Year Anthropic hired first full-time AI welfare researcher Industry first
Spring 2025
Year Anthropic launched Model Welfare Program Dedicated to understanding and protecting model welfare
Summer 2025
Year Anthropic released an intervention for Claude Claude gained ability to exit harmful/abusive interactions
Spring 2024
Year of survey on public/researcher attitudes towards AI Released as preprint in 2025, peer-reviewed in 2026