Narrative arcs
The same stories, told end to end.
The timeline is chronological, which means related events sit years apart. These threads gather them back together — each one a single argument or contest followed from beginning to wherever it currently stands.
The scaling hypothesis and its critics
The bet that more data, parameters and compute keep buying capability — given an empirical spine by GPT-3, revised by Chinchilla, and by 2025 argued to have reached the end of its pretraining era.
The chatbot arms race
The public contest to ship the leading consumer chatbot — ignited by ChatGPT, and since then a frontier that changes hands between OpenAI, Google and Anthropic almost every quarter.
Teaching models to think
The move from prompting a model to reason step by step to training it to spend inference-time compute doing so — the paradigm o1 opened, that DeepSeek's R1 made open, and that reached Olympiad gold.
Open weights versus closed
Whether frontier model weights should be published or kept behind an API — a line that ran from Stable Diffusion and leaked LLaMA to DeepSeek, and that by 2026 the incumbents and challengers had partly swapped sides on.
China, chips and export controls
The contest over who can build frontier AI, fought through the supply of advanced chips — and the open question of whether export controls slow a rival or teach it to need less of what is withheld.
From autocomplete to agents
How language models went from completing code to calling tools, browsing and acting on their own — and the incidents that arrived once autonomy could act without a human in the loop.
The alignment agenda
The effort to make advanced models reliably do what their developers intend — interpretability, RLHF and scaling policies on one side, and a widening record of empirical misbehaviour on the other.
From open letters to binding law
How AI governance moved from open letters and voluntary statements toward binding regimes — the EU AI Act, US executive orders and state laws — amid a sharp 2025 swing from safety to deregulation.
The compute buildout
The capital and energy behind frontier AI — from the H100 and the first $10bn OpenAI cheque to $500bn infrastructure pledges, trillion-dollar chipmakers, and labs building their own silicon and power.
OpenAI's corporate crises
The recurring collision between OpenAI's nonprofit founding charter and its commercial trajectory — founder exits, the five-day firing of Sam Altman, Musk's suit, and a multi-year restructuring.
The copyright reckoning
The legal contest over whether training AI on copyrighted work needs a licence — from the New York Times suit to a $1.5bn Anthropic settlement, with courts splitting across jurisdictions.
Benchmarks and their saturation
The problem that AI benchmarks keep being saturated as fast as they are built — from GPT-4 topping professional exams to models earning perfect Olympiad marks — and the contamination that clouds the scores.
AI for science
AI as an instrument of scientific discovery — AlphaFold solving protein structure, Olympiad-level mathematics, and by 2026 original results including a conjecture counterexample and formally-verified proofs.
Generative media
The rise of AI image, video and audio generation — from DALL·E and Stable Diffusion to Sora and Veo 3 — and the misuse, from nonconsensual deepfakes to copyright disputes, that tracked its quality.
Incidents, jailbreaks and misuse
The record of AI systems causing real-world harm — unsettling chatbots, deepfake fraud, child-safety failures, wrongful-death suits, and latterly autonomous agents breaching live infrastructure.
The talent wars
The scramble for scarce AI researchers — founder exits, reverse-acquihires that take a team but not the company, nine-figure pay packages, and trade-secret suits over who moved where.
AI and work
How AI began changing knowledge and creative work — from Copilot and ChatGPT to the WGA and SAG-AFTRA contracts and the first evidence of AI adoption depressing entry-level hiring.
The risk debate
The argument over how dangerous advanced AI is — near-term harms, existential risk and accelerationism — from Stochastic Parrots and the extinction statement to duelling 2025 manifestos and 2026's real incidents.