How to use your time and money for good — as effectively as possible — with Will MacAskill

April 25, 2024

46 min episode · 2 min read

Will Macaskill

Episode

46 min

Read time

2 min

AI-Generated Summary

Published Jan 11, 2026

Key Takeaways

✓Three-Question Framework: Evaluate charitable causes by asking which problems are biggest in scale, most tractable when effort is applied, and most neglected by current resources to identify where donations create outsized impact beyond personal connections or emotional appeals.
✓Giving Pledge Structure: Commit to donating the higher of either 10% of annual income or 2.5% of net worth. If one-third of capable donors adopted this, it would generate $3-4 trillion annually, potentially eliminating philanthropy bottlenecks and fulfilling reasonable moral obligations.
✓Tool AI vs Agent AI: Accelerate development of tool AI systems that provide input-output responses like ChatGPT while heavily regulating agentic AI that acts autonomously in the world. This distinction helps capture AI benefits while managing existential risks from systems operating independently.
✓EA as Process Not Prescription: Effective altruism functions like science—a method of using evidence and reasoning to maximize good, not a fixed set of recommended charities. Practitioners expect to change their minds repeatedly as new evidence emerges, maintaining intellectual humility throughout.

What It Covers

Will MacAskill discusses effective altruism principles, addressing the Sam Bankman-Fried scandal, defending EA as a process not prescription, and exploring AI risks, giving pledges, and maximizing philanthropic impact through evidence-based approaches.

Key Questions Answered

•Three-Question Framework: Evaluate charitable causes by asking which problems are biggest in scale, most tractable when effort is applied, and most neglected by current resources to identify where donations create outsized impact beyond personal connections or emotional appeals.
•Giving Pledge Structure: Commit to donating the higher of either 10% of annual income or 2.5% of net worth. If one-third of capable donors adopted this, it would generate $3-4 trillion annually, potentially eliminating philanthropy bottlenecks and fulfilling reasonable moral obligations.
•Tool AI vs Agent AI: Accelerate development of tool AI systems that provide input-output responses like ChatGPT while heavily regulating agentic AI that acts autonomously in the world. This distinction helps capture AI benefits while managing existential risks from systems operating independently.
•EA as Process Not Prescription: Effective altruism functions like science—a method of using evidence and reasoning to maximize good, not a fixed set of recommended charities. Practitioners expect to change their minds repeatedly as new evidence emerges, maintaining intellectual humility throughout.

Notable Moment

MacAskill describes looking at photos of children with debilitating diseases like elephantiasis to test whether he could justify keeping income versus donating half his earnings, concluding one prevented case would justify the sacrifice multiplied by hundreds of actual impacts.

Know someone who'd find this useful?