AI Reading List

Vishal Maini
Jun 19 · 5 min read

For newcomers to the field of artificial intelligence, prioritizing among endless AI resources can be an overwhelming challenge. This list attempts to do exactly that: it’s a carefully curated compilation of resources for getting up to speed quickly on key topics in artificial intelligence research and its long-term implications.

The list is divided into “” sections with a few high-priority readings, for maximum value with minimal time investment, and “deep dive” sections for further exploration.

Readers need not be technical, nor have a prior background in artificial intelligence. The list may be of special interest to those considering entering the field of AI research or adjacent fields, whether in technical or non-technical roles.

This list covers high-level context (“What is intelligence, and what would it mean to re-create it in machines?”) technical foundations (“How does narrow AI work today, and what are some of the favored technical approaches towards general AI?”), safety considerations (“What will it take for beyond-human-level AI to be safe and act in accordance with human preferences?”), and strategic questions (“How can we coordinate towards beneficial outcomes from advanced AI?”).
To get the most value out of these resources, consider actively reading. Write short summaries of key concepts and takeaways that you can refer to later (see: ). Spend some time critically analyzing the ideas. What makes intuitive sense to you, and what doesn’t? What are your criticisms of the ideas presented? How do concepts fit together?

1. Intelligence

What is intelligence, and how might we re-create it in machines? Why now? Three ingredients for AI progress — compute, data, and algorithms.

80/20

Deep dive

2. Machine Learning

Enabling machines to learn for themselves. Learning to make predictions and identify patterns given different kinds of data (supervised, unsupervised, and reinforcement learning). Demystifying the objective function.

80/20

Deep dive

3. Deep Learning

Learning to predict an output given an input; drawing inspiration from the brain. Looking inside the black box of deep neural networks. What is a cat, and why? How do we represent the world around us numerically (as sensory “input” or data) and use math to make sense of it? Common architectures.

80/20

Deep dive

4. Deep Reinforcement Learning

Artificial agents learning from reward. Value functions. Exploration and exploitation. Reaching superhuman performance in complex games. Key breakthroughs.

80/20

Deep dive

5. Safety & Alignment

How do we ensure that highly capable and general AI systems reliably understand what we want and help us get it? Identifying and pre-empting potential failure modes. An introduction to specification, robustness, and assurance.

80/20

Deep dive

6. Strategy & Governance

Near- and long-term strategic challenges and opportunities presented by AI. An introduction to the AI governance problem: the problem of devising global norms, policies, and institutions to best ensure the beneficial development and use of advanced AI.

80/20

Deep dive

You can also keep up to date with the latest developments in the AI space by signing up for by Jack Clark and the by Rohin Shah, and reading them closely every week.

If you enjoyed these resources and are interested in working on the challenges and opportunities presented by artificial intelligence research, check out the 80,000 Hours to see who’s hiring. If you have questions or feedback, feel free to .

Thanks to , , , , , , , , , , , , and for contributing resources and providing feedback on earlier versions of this list.

Machine Learning for Humans

Demystifying artificial intelligence & machine learning. Discussions on safe and intentional application of AI for positive social impact.

Vishal Maini

Written by

Strategy & communications @DeepMindAI. Previously @Upstart, @Yale, @TrueVenturesTEC. Views expressed here are my own.

Machine Learning for Humans

Demystifying artificial intelligence & machine learning. Discussions on safe and intentional application of AI for positive social impact.