AI Safety 101

Semua kursus

• LA

  AI Risk Fundamentals

  Lens Academy

  Based heavily on the book "If Anyone Builds It, Everyone Dies" – covering the core case for why superintelligent AI could be catastrophic, and what should be done about it.

  Buka kursus
• CF

  AI Safety, Ethics and Society (AISES)

  Center for AI Safety (CAIS)

  Covers a wide range of risks, including loss of control, misalignment, malicious use, etc, while leveraging concepts and frameworks from existing research fields to analyze AI safety.

  Buka kursus
• GD

  DeepMind AGI Safety Short Course

  Google DeepMind

  75-minute course covering alignment problems expected as AI capabilities advance. Includes recorded talks, exercises, and an accompanying workbook.

  Buka kursus
• CL

  Existential Risks Introductory Course (ERIC)

  Cambridge Long View Initiative (CLVI)

  Each section covers a different topic related to existential risks. The book 'The Precipice' by Toby Ord forms the majority of the core reading material.

  Buka kursus
• BI

  Future of AI

  BlueDot Impact

  Very short, engaging introduction to AI and its potential impacts. Good entry point to share with laypeople curious about AI.

  Buka kursus
• BI

  Intro to Transformative AI

  BlueDot Impact

  Intensive 5-unit course helping people fast-track their understanding of AI’s impact on society and start taking their first steps in making a contribution to AI safety.

  Buka kursus
• SU

  Introduction to AI Safety

  Stanford University

  What is safe AI, and how do we make it? This Stanford University course explores that question, focusing on the technical challenges of creating reliable, ethical, and aligned AI systems.

  Buka kursus
• CA

  Introduction to Cooperative AI

  Cooperative AI Foundation (CAIF)

  A primer on cooperative AI, a research field focused on improving the cooperative intelligence of advanced AI for the benefit of all.

  Buka kursus
• EK

  Reading What We Can

  Esben Kran

  Collection of books and articles for a 20 day reading challenge. Covers AI safety basics, ML engineering, ML upskilling, and AI safety-relevant sci-fi.

  Buka kursus
• EN

  Deep Dive

  European Network for AI Safety (ENAIS) and AI Safety Hungary

  Designed as a 201 AI policy course, helping you find your footing in the vast AI governance landscape by engaging with various dimensions.

  Buka kursus
• BI

  AGI Strategy

  BlueDot Impact

  The decisions being made today determine whether AI liberates humanity, destabilizes it, or worse. This course aims to prepare participants to be part of those decisions.

  Buka kursus
• RR

  AI Futurism Reading List

  Redwood Research

  Guide to readings which seem valuable for thinking about key future dynamics in advanced AI development – timelines, takeoff, misalignment, control, governance, and strategy.

  Buka kursus
• BI

  Economics of Transformative AI

  BlueDot Impact

  Course designed for economists who want to develop their understanding of transformative AI and its economic impacts.

  Buka kursus
• AE

  Agent Foundations for Superintelligence-Robust Alignment

  Alignment Ecosystem Development (AED)

  Guide to the cluster of thought which expects solving alignment in a way which scales to superintelligence to require well-specified proposals guided by strong theoretical understanding.

  Buka kursus
• V

  AI Alignment Forum: Curated Sequences

  Various

  List of sequences curated by the AI Alignment Forum team, featuring work from Richard Ngo, Paul Christiano, etc.

  Buka kursus
• CP

  AI Safety Atlas

  Centre pour la Sécurité de l'IA (CeSIA)

  Textbook systematically mapping the landscape of AI safety knowledge. Ideas are laid out in a carefully structured narrative where concepts build naturally on previous ones.

  Buka kursus
• A

  Alignment Research Engineer Accelerator (ARENA)

  ARENA

  Facilitates upskilling in machine learning engineering for the purpose of contributing directly to AI alignment in technical roles.

  Buka kursus
• BI

  BlueDot Impact: Technical & Frontier AI Governance

  BlueDot Impact

  Covers key concepts and research perspectives in AI safety, split into two main streams: Technical AI Safety and AI Governance.

  Buka kursus
• CF

  CHAI: Annotated Bibliography of Recommended Materials

  Center for Human-Compatible AI (CHAI)

  Reading list discussing ideas CHAI believes warrant further attention and research for the purpose of developing human-compatible AI.

  Buka kursus
• BB

  CS 2881: AI Safety

  Boaz Barak

  Graduate-level course at Harvard University examining technical challenges in AI safety, including adversarial robustness, jailbreaks, interpretability, and potential catastrophic capabilities.

  Buka kursus
• I

  Iliad Intensive

  Iliad

  In-depth introduction to foundational AI alignment for people with strong backgrounds in mathematics, physics, or theoretical computer science.

  Buka kursus
• CF

  Introduction to ML Safety

  Center for AI Safety (CAIS)

  PhD-level survey course on technical ML safety. Covers various technical topics to reduce existential risks from AI, including robustness, monitoring, control, and systemic safety.

  Buka kursus
• PO

  Key Phenomena in AI Risk

  Principles of Intelligence

  Focused on introducing people to AI risk factors from misdirected optimization or consequentialist cognition, aiming to remain largely agnostic of solution paradigms.

  Buka kursus
• MA

  Language Models and Intelligent Agentic Systems

  Meridian and C2D3

  Series of 16 lectures explaining how language model systems are built, in order to understand and predict their behavior.

  Buka kursus
• VK

  Learning-Theoretic Agenda Reading List

  Vanessa Kosoy

  Vanessa Kosoy's math-heavy reading list for people interested in studying the learning-theoretic agenda.

  Buka kursus
• MA

  MATS AI Safety Strategy Curriculum

  ML Alignment & Theory Scholars (MATS)

  As part of their program, MATS runs a series of discussion groups focused on topics relevant to prioritizing research in AI safety. This is the curriculum.

  Buka kursus
• UE

  Tracks @ XLab

  UChicago Existential Risk Laboratory (XLab)

  A technical track on AI control and a governance track on verifying international AI agreements – with primary literature, interactive demos, and writing practice.

  Buka kursus
• UE

  UChicago XLab AI Security Guide

  UChicago Existential Risk Laboratory (XLab)

  Introduces AI security research through five sections covering adversarial examples, jailbreaking, model tampering, data poisoning, and information extraction.

  Buka kursus
• VK

  Victoria Krakovna: AI Alignment Resources

  Victoria Krakovna

  Intended to help people get up to speed on the main ideas in AI alignment i.e. ensuring advanced AI systems do what we want them to do.

  Buka kursus

Artikel penjelas dari Tim MAI, dalam bahasa Indonesia, segera hadir.