Start here · choose four sources or the full three hours
AI is the biggest story of the decade, and a surprising amount of the public argument is still people summarising things they have not read.
Three hours will not make you an expert. Expertise comes from doing the work, and reading is not the work. What this does is more modest and more useful: it gets you to where you can follow the conversation, check claims against primary sources, and notice when someone is bluffing. Nine tracks after this one. Every entry says what the source actually argues before you click it. Links go to real sources, never to searches.
118 sources · 10 tracks. 107 are free outright. 7 are books, free with a library card. 3 are films you rent or stream. 1 is a paid book, and optional.
Choose the useful amount
Do not start with vibes. Start with the machine doing the trick, because once the trick is visible, half the discourse gets less mystical and the other half gets more interesting. This is the warm-up lap, not the monastery.
The best short explanation of transformer mechanics in any medium: embeddings as directions in space, attention as tokens updating each other's meaning, softmax at the end. You will not be able to implement one afterward, and you will understand every subsequent conversation better.
Also in Technical foundations. Ticking it here marks it there.
27 min YouTube · free
Ready with the full reading map.
Aggregate usage diagnostics are stored; your question and answer text are not.
Capabilities, costs and trend lines before prophecy. The important habit is boring in the best way: look at the live measurements, notice who made them, and keep one hand on your wallet when a chart arrives wearing a cape.
A broad public data layer for the whole subject: investment, compute, model capability, adoption, public opinion and timelines, with reusable charts and citations. It is less inside-baseball than the lab documents and therefore useful early, before the acronyms begin breeding in the walls.
20 min ourworldindata.org · free
The field's annual census: capability, investment, adoption, cost, policy and public opinion, all sourced. Skim the top takeaways on release, then use it year-round as the reference whenever someone quotes a number at you. The adoption-versus-measured-value gap in the economy chapter is the most interesting figure in it and the one most often skipped.
Also in Frontier tracking, Economics & infrastructure, Applications, Keep current. Ticking it here marks it there.
20 min – 2 hr hai.stanford.edu · free
Training compute, hardware price-performance, algorithmic efficiency, inference cost and data supply, as live charts. Two facts do most of the work elsewhere: frontier compute has grown several-fold annually for a decade, and the cost of any given capability falls fast once it exists.
Also in Economics & infrastructure. Ticking it here marks it there.
15 min epoch.ai · free
Abandons benchmark scores for task duration: which lengths of human-professional work a model finishes at 50% reliability. That horizon has doubled roughly every seven months for six years, through several architecture changes. The caveat almost every citation drops is that the 80% reliability curve sits about two doublings, or a year, behind.
Also in Evals & benchmarks, The risk argument. Ticking it here marks it there.
30 min metr.org · free
An attempt to measure economically valuable work instead of exam-shaped cleverness: realistic deliverables across 44 knowledge-work occupations, judged against expert outputs. The limitations matter as much as the scores, because the real world cruelly insists on iteration, ambiguity and people changing their minds after lunch.
Also in Evals & benchmarks, Economics & infrastructure, Applications. Ticking it here marks it there.
20 min openai.com · free
This is the compressed version of the whole fight: institutional middle, maximal alarm, serious skepticism. Read all three before deciding you have found the grown-up in the room. The room is mostly grown-ups disagreeing at high speed.
Over a hundred experts nominated by thirty-plus countries plus the EU, UN and OECD, with Key Updates through the year when capabilities move. Its most valuable feature is structural: it separates established from contested from speculated and refuses to collapse the third into the first. Read the four-page executive summary, then audit anything you believe confidently against which category it falls in.
Also in The risk argument, Keep current. Ticking it here marks it there.
25 min – 3 hr internationalaisafetyreport.org · free
The maximalist case compressed past politeness: we do not know how to give a system any particular goal, we get one attempt, and the failure mode is everyone dying. Argues the current paradigm cannot produce a mind that likes us because we can neither specify nor inspect what we are building. Calls for an indefinite worldwide moratorium enforced by international agreement.
Also in The risk argument. Ticking it here marks it there.
12 min TED · free
Separates AI methods from applications from adoption. Methods can improve exponentially while the other two move at institutional speed, and historically always have: safety-critical domains resist, regulation binds, liability attaches, infrastructure lags. Concludes that superintelligence discourse mistakes capability for power, since power requires the world to reorganise around you and the world is slow and full of humans with lawyers. Read part one for the briefing; the whole thing later.
Also in The risk argument. Ticking it here marks it there.
15 min – 1.5 hr Knight First Amendment Institute · free