AI Research
AI Research
The papers, ideas, and discoveries that actually moved the needle. Not a comprehensive literature review — a curated collection of the research that changed something: launched an era, proved a thesis, or opened a new direction.
Each paper here is annotated with why it matters, who wrote it, and what it connects to. The goal is understanding, not citation counting.
Landmark Papers
The ones that changed the trajectory of the field:
- Attention Is All You Need (2017) — Introduced the Transformer. The architecture behind everything that followed. Google Brain.
- Scaling Laws for Neural Language Models (2020) — Proved that AI performance scales predictably with compute. Gave the industry its roadmap. Jared Kaplan, Johns Hopkins University / OpenAI.
- Constitutional AI (2022) — Anthropic‘s approach to alignment: principles over preferences. Alternative to RLHF.
More to come: GPT-3, BERT, AlphaFold, Chinchilla, InstructGPT, DPO…
Understanding AI
Not papers, but explanations. How things actually work inside these systems.
- How LLMs Work — The full pipeline from text in to text out. Start here if you want the big picture.
Models & Architectures
Deep dives into specific architectures, how they differ, and what trade-offs they make.
models/— To be populated (Transformer variants, MoE, State Space Models, etc.)
How to Read a Paper (for non-academics)
If you’re new to research papers, here’s the shortcut:
- Read the abstract — 1 paragraph summary
- Look at the figures/tables — They usually tell the whole story
- Read the introduction — Why the work was done
- Read the conclusion — What they found
- Skip the methodology (unless you’re implementing it)
- Check who cited it — Impact is measured by what it enabled
Or just read our annotated summaries. That’s what they’re for.
Go Deeper
- AI Learning Centre — If research feels too advanced, start with structured learning
- Heroes of AI — The people writing these papers
- AI Companies — The labs funding this research
- AI Timeline — Where these papers sit in history
- AI Intelligence Hub — Back to the hub home