Learning About AI
These are videos, newsletters, podcasts and courses for students who want to understand how AI and large language models work, use them well, and follow the debates about them. If you are new to the topic, begin with Start here and Basic AI literacy. After that, the list is sorted into three levels:
- Non-technical: no math and no code. What AI is, what it does to society, and the news.
- Technical: how it works under the hood. Some math or coding helps, but these are made for learners.
- Pro level: long, deep or insider material for students who want to build AI or follow the field closely.
Every source has a point of view. Some people think AI could be dangerous on a civilizational scale. Others think that worry is overblown, and still others mainly see a business opportunity. Some sources are made by AI companies or by people who invest in them. The Lens notes say where each source is coming from, so you can read it with that in mind.
Start here
Timothy B. Lee & Sean Trott, "Large language models, explained with a minimum of math and jargon" (2023). Word vectors, attention and training in plain English.
Lens Lee is a longtime tech journalist (Ars Technica, Vox, Washington Post); Trott is a cognitive scientist at UC San Diego. Careful and non-hyped.
Andrej Karpathy, "Deep Dive into LLMs like ChatGPT" (2025, 3.5 hours). How a chatbot is built, from raw internet text to the finished assistant. No math needed.
Lens A founding member of OpenAI and former head of AI at Tesla, now mostly a teacher. An insider, enthusiastic about the technology but frank about its quirks.
3Blue1Brown, "Neural Networks" series (Grant Sanderson). Animated lessons on how neural nets learn, ending with chapters on GPT and attention.
Lens A math educator with no stake in the AI industry.
Basic AI literacy
What every student should know before using AI for school or work. No math, no code.
What an AI chatbot is
Financial Times, "Generative AI exists because of the transformer" (2023). Short visual story on how chatbots predict the next word.
Elements of AI (University of Helsinki). Free self-paced course on what AI is and what it can and can't do.
Lens A public-education project; neutral.
AI for Everyone (Andrew Ng, DeepLearning.AI). Short non-technical course; free to audit.
Lens Ng co-founded Google Brain and Coursera. Very optimistic about AI and openly dismissive of fears that it could wipe out humanity.
Using AI well
Anthropic, "AI Fluency: Framework and Foundations." Free course on when to use AI, how to ask, and how to check the answer.
Lens Made by the company that sells Claude.
OpenAI Academy. Free short videos and guides on using ChatGPT.
Lens Made by the company that sells ChatGPT.
Ethan Mollick, Co-Intelligence: Living and Working with AI (2024). Short book on using AI as a helper rather than a replacement.
One Useful Thing (Ethan Mollick). Posts on how to use the newest tools, usually within days of release.
Lens A Wharton management professor who studies AI at work. Enthusiastic and practical; critics say he can be too upbeat about new models.
Limits and risks: errors, bias, flattery, hype
Carl Bergstrom & Jevin West, "Modern-Day Oracles or Bullshit Machines?" Free online course on thinking critically about chatbots, written for students.
Lens University of Washington professors and authors of Calling Bullshit. Skeptical, but they teach you to judge for yourself.
Arvind Narayanan & Sayash Kapoor, AI Snake Oil (2024). How to tell AI that works from AI that doesn't. Their blog is in the policy essentials below.
Andy Masley. What AI actually costs in energy and water, with numbers.
Lens Argues that one person's chatbot use is a tiny part of their environmental footprint, and pushes back on popular claims that it's large. Other researchers put more weight on the overall growth of data centers.
Essentials for keeping up with AI policy debates
The Ezra Klein Show (NYT). Long conversations on what AI means for politics, work and society, with lab leaders, critics and policymakers.
Lens Klein is a liberal NYT columnist, co-founder of Vox, and co-author of Abundance. He takes the idea that AI will be transformative seriously and hosts people from all sides of the debate.
Also on: YouTube
Dwarkesh Podcast (Dwarkesh Patel). Long interviews with AI researchers, lab leaders and forecasters. Podcast, YouTube and newsletter.
Lens An independent interviewer known for well-prepared, technical questions. Most guests expect very powerful AI soon, so the show leans toward that view.
Also on: YouTube
Don't Worry About the Vase (Zvi Mowshowitz). Weekly roundups of AI news, research and policy fights. Very long; skim the headings.
Lens A writer from the "rationalist" online community and a former professional Magic: The Gathering player. He is seriously worried that advanced AI could cause a catastrophe, and his commentary is opinionated.
80,000 Hours Podcast. Interviews on AI risk, governance and careers in AI policy.
Lens A career-advice nonprofit that grew out of the effective altruism movement. It ranks risks from advanced AI as the world's most pressing problem, and many guests are deeply concerned about AI safety.
Also on: YouTube
Justified Posteriors: The Economics of AI (economists Seth Benzell & Andrey Fradkin). What the research says about AI and jobs, growth and markets. Podcast and newsletter.
Lens Academic economists. They state their beliefs before reading a paper and then say how much it changed their minds. Data-driven and fairly neutral.
Also on: YouTube
The AI Policy Podcast (CSIS, Center for Strategic and International Studies). US AI policy, export controls and national security.
Lens A bipartisan Washington think tank with a national-security framing. Much of it is about competition with China.
Also on: YouTube
AI as Normal Technology (Arvind Narayanan & Sayash Kapoor, Princeton). AI as a powerful but ordinary technology that will spread slowly, not a coming superintelligence. Formerly the AI Snake Oil blog.
Lens Computer scientists; Narayanan directs Princeton's Center for Information Technology Policy. Skeptical of both the hype and the doom, and a useful counterweight to 80,000 Hours and Zvi.
Non-technical
Books
Melanie Mitchell, Artificial Intelligence: A Guide for Thinking Humans (2019). Written before ChatGPT, but still one of the clearest books on what "intelligence" means and where AI falls short.
Newsletters
Best bets
Understanding AI (Timothy B. Lee). Explains each week's AI news without hype.
Lens See Lee above; a journalist with a computer science background.
AI: A Guide for Thinking Humans (Melanie Mitchell).
Lens A computer scientist at the Santa Fe Institute. Skeptical of claims that chatbots truly "understand" or reason like people.
AI and politics, policy, society
Programmable Mutter (Henry Farrell). AI, technology and political economy.
Lens A political scientist at Johns Hopkins. He argues chatbots are best understood as social and cultural technologies, like markets or libraries, rather than as artificial minds.
Hyperdimensional (Dean Ball). US AI policy.
Lens Worked on AI policy in the Trump White House in 2025. Libertarian-leaning and wary of heavy regulation, but takes AI risks seriously. A good right-of-center policy voice.
AI Policy Perspectives. Short pieces on AI governance from many writers.
Lens Run by Google DeepMind's policy research team, so it comes from inside an AI company.
AI Frontiers. Experts debating AI's effects on security, work and society.
Lens Backed by the Center for AI Safety. It publishes a range of views, but its backers are focused on AI risk.
Luiza's Newsletter (Luiza Jarovsky). AI law and regulation, including the EU AI Act.
Lens A privacy and tech lawyer who also sells AI governance training. Pro-regulation and European in outlook.
Never Met a Science (Kevin Munger). What the internet and AI do to politics and society.
Lens A political scientist who writes provocatively and openly speculates.
Pixels and Patterns (Steven Denney & Aron van de Pol). Research notes on social science, digital humanities and AI.
Lens Two Leiden University researchers sharing work in progress.
Asterisk Magazine. Long essays, many on AI.
Lens Comes out of the effective altruism community, so it often takes AI risk seriously, but it publishes a range of views.
Cosmos Institute. AI and philosophy.
Lens A philosophy-focused nonprofit with a classical-liberal outlook. It focuses on protecting human freedom and judgment as AI spreads.
AI safety and the future (one side of the debate; say so)
Dario Amodei (Anthropic CEO). Long essays, such as "Machines of Loving Grace."
Lens Runs one of the leading AI companies. His essays mix big optimism with warnings about risk, and his company profits from AI.
AI Futures Project. The team behind the "AI 2027" scenario.
Lens Led by Daniel Kokotajlo, a former OpenAI researcher who left over safety concerns. They expect superhuman AI within years and publish updates as their forecasts change.
AI Safety Newsletter (Center for AI Safety).
Lens The group behind the 2023 statement that the risk of extinction from AI should be a global priority.
Clear-Eyed AI (Steven Adler).
Lens A former OpenAI safety researcher who is now a critic of how AI labs handle safety.
Lens An advocacy group campaigning to stop the development of superintelligent AI. Activism, not analysis.
AI in education
AI + Education = Simplified (Lance Eaton).
Lens A higher-education teaching and learning specialist; practical and generally open to AI in the classroom.
AI, academia, and the Future (Bryan Alexander). AI and universities.
Lens A futurist who writes about where higher education is heading.
No More Marking. AI and assessment in schools.
Lens A UK company that sells school assessment tools, so it writes partly about its own products.
AI at work
Nate B Jones. Daily posts and videos on AI for working professionals.
Lens Aimed at executives and business users; high volume and business-focused.
Also on: YouTube · Apple Podcasts
YouTube
News explained
AI Explained. Breakdowns of new models and papers, with attention to what the benchmarks do and don't show.
Lens An independent creator who built his own reasoning test (SimpleBench). Measured, and skeptical of overblown claims.
Also on: Apple Podcasts
Cleo Abram (Huge If True). Tech explainers for a general audience.
Lens A former Vox video journalist whose show is deliberately optimistic about technology.
Bloomberg Tech. Tech and business news.
Lens Covers AI as business and markets.
Also on: Apple Podcasts
WIRED. Tech news and features.
Lens Often critical of Silicon Valley power. Its podcast, Uncanny Valley, is listed under Podcasts.
Interviews, debates, documentaries
Species | Documenting AGI. Documentary-style videos.
Lens Makes the case that advanced AI could threaten humanity. Persuasive rather than balanced.
Intelligence Squared. Formal debates, several on AI.
Lens A London-based debate series; you hear both sides.
Also on: Apple Podcasts
Big Think. Short talks by experts.
Lens Quality varies a lot by speaker.
Also on: Apple Podcasts
Straight from the AI companies (useful, but it's their marketing)
Anthropic · Claude · OpenAI · Google DeepMind · AI at Meta
Podcasts
Best bets
AI Summer (Timothy B. Lee). Interviews with experts on AI technology and policy. Also a newsletter.
Lens See Lee above.
Machine Gods (NPR). The people and beliefs behind AI.
Lens Public radio journalism about the beliefs and personalities driving the AI race; reported, not advocacy.
Also on: YouTube
The Last Invention (Longview; Gregory Warner & Andy Mills). Narrative series on the race to build AI and the fights over it.
Lens Made by former NPR and New York Times journalists. It gives doomers, accelerationists and skeptics each a hearing.
Also on: YouTube
Google DeepMind: The Podcast (Hannah Fry). DeepMind researchers explain their work.
Lens Fry is a mathematician and broadcaster, but this is Google's own show.
Also on: YouTube
Uncanny Valley (WIRED). Weekly tech news.
Lens WIRED journalists; often critical of big tech.
Also on: YouTube
Search Engine (PJ Vogt). Questions about tech and the internet; some AI episodes.
Lens A curious general-audience reporter, formerly of Reply All.
Interviews
Big Technology Podcast (Alex Kantrowitz). Interviews with tech executives and reporters, plus weekly news.
Lens A former BuzzFeed News tech reporter who interviews tech executives. Good access, but interviews with CEOs can be gentle.
Also on: YouTube
Core Memory (Ashlee Vance). Podcast and newsletter with reported stories on the tech and AI industry.
Lens A former Bloomberg Businessweek reporter and Elon Musk biographer. Admires engineers and founders.
Also on: YouTube · Newsletter
AI & I (Dan Shipper). How people use AI at work.
Lens Shipper runs Every, a media company that also builds AI products. Enthusiastic.
FT Tech Tonic (Financial Times).
Lens Business journalists; skeptical of hype and focused on money and power.
Plain English with Derek Thompson. Some AI and economy episodes.
Lens A center-left writer, formerly of The Atlantic, and Ezra Klein's co-author on Abundance.
Also on: YouTube
Interesting Times (Ross Douthat). Some AI episodes.
Lens A conservative, Catholic NYT columnist who asks what AI means for religion, meaning and human nature.
Also on: YouTube
All-In (Chamath Palihapitiya, Jason Calacanis, David Sacks, David Friedberg). Tech investors on AI, markets and politics.
Lens A US right / Silicon Valley view. The hosts invest in tech, and Sacks served as Trump's White House AI and crypto czar.
Also on: YouTube
Education
AI in Education Podcast (Dan Bowen & Ray Fleming).
Lens Education-technology insiders based in Australia; practical and positive about AI in schools.
Also on: YouTube
Data and trackers (to check claims yourself)
Stanford AI Index. Yearly report with charts on AI research, cost, use and policy.
Lens From Stanford's Institute for Human-Centered AI; widely cited by both sides.
Our World in Data: AI. Charts on AI progress and use.
Lens An Oxford-based nonprofit focused on data; neutral.
Arena (formerly LMArena / Chatbot Arena). Public leaderboard of chatbots, ranked by people's votes in blind comparisons.
Lens Started by UC Berkeley researchers, now a company. A 2025 study argued that big AI labs can game the rankings.
Technical
YouTube
3Blue1Brown. Animated math: neural nets, gradient descent, transformers, attention.
StatQuest with Josh Starmer. Stats and machine learning in small steps. Has full neural network and transformer playlists.
Lens A former genetics researcher at UNC who teaches stats; no industry stake.
Welch Labs. Long videos on how neural nets learn, and on AI history.
IBM Technology. Short whiteboard explainers: what is RAG, what is an AI agent, what is fine-tuning.
Lens A company channel, but mostly neutral explainers rather than ads.
Two Minute Papers (Károly Zsolnai-Fehér). Short summaries of new AI research papers.
Lens A researcher who is excited about every result. Enjoy it, but take the tone with a grain of salt.
Grant Sanderson. His personal channel; extra talks, including one on how transformers work.
freeCodeCamp.org. Long, free coding courses: Python, machine learning, building with LLM APIs.
Lens A nonprofit.
Read and click through
Transformer Explainer (Georgia Tech). Click through a live GPT-2 model in your browser.
LLM Visualization (Brendan Bycroft). 3D walk-through of every step inside a small GPT.
Jay Alammar, "The Illustrated Transformer" (2018). The standard picture-based explainer of the transformer.
Stephen Wolfram, "What Is ChatGPT Doing … and Why Does It Work?" (2023). Long essay, still accurate on the basics.
Lens The founder of Wolfram Research (Mathematica); brilliant, and fond of tying things back to his own work.
Anthropic, "Tracing the thoughts of a large language model" (2025). What researchers find when they look inside a model.
Lens Research from an AI company about its own model.
Free courses (need some coding)
fast.ai, Practical Deep Learning for Coders (Jeremy Howard).
Lens An independent course that champions making AI accessible to non-experts.
Lens From the company that hosts most open-source AI models; pro open source.
Newsletters
Simon Willison's Newsletter. Hands-on notes on new models and tools.
Lens Co-creator of the Django web framework and the person who coined "prompt injection." A heavy user who is blunt about flaws and security risks.
Import AI (Jack Clark). Weekly digest of AI research papers.
Lens A former Bloomberg reporter, now an Anthropic co-founder who works on policy. Thoughtful, but he works for an AI lab and worries a lot about risk.
Data
Epoch AI. Data on training compute, model size and trends.
Lens An independent research group whose numbers are widely cited by journalists and policymakers.
Pro level
Courses (serious time and math)
Andrej Karpathy, "Neural Networks: Zero to Hero." Build a GPT from scratch in Python.
Stanford Online. Full Stanford courses: CS229 (Machine Learning), CS224N (NLP with Deep Learning), CS25 (Transformers United).
MIT OpenCourseWare. Full MIT courses, including AI and machine learning.
YouTube
Machine Learning Street Talk. Long technical interviews with AI researchers.
Lens Leans philosophical, and often hosts researchers who doubt that today's chatbots are on a path to human-level intelligence.
Also on: Apple Podcasts
AI Engineer. Conference talks from people building AI products. For developers.
Lens Industry practitioners; often vendors showing their tools.
Newsletters
Interconnects (Nathan Lambert). How models are trained, especially open models.
Lens An AI researcher who works on training methods and argues strongly for open-source AI, especially in the US.
Strange Loop Canon (Rohit Krishnan). Essays on AI and economics.
Lens An investor and engineer; optimistic and idea-driven.
SemiAnalysis. Chips, data centers and the AI supply chain. Mostly paywalled.
Lens Industry analysts who write for investors and companies.
Startup / investor view (insider, hype-heavy)
Lens for this group: investors and boosters who profit when AI succeeds.
Y Combinator. The startup accelerator; talks and its Startup Podcast.
Also on: Apple Podcasts
a16z (Andreessen Horowitz). Venture firm videos and podcast.
Lens Has lobbied against AI safety rules, such as California's SB 1047.
Also on: Apple Podcasts
TBPN. Daily live tech talk show.
Also on: Apple Podcasts
No Priors (Sarah Guo & Elad Gil). Venture investors talking to founders and researchers.
Also on: Apple Podcasts