About this project

The AI History Project is an interactive chronicle of artificial intelligence — the concepts, people, and organizations that shaped the field, decade by decade, from the 1940s to today. Alongside the timeline you'll find data visualizations and a knowledge graph connecting every entry — with a time scrubber, a semantic constellation view, and an AI-powered question box.

Educational and experimental

This site is an educational resource: a structured way to see how eight decades of ideas, breakthroughs, and institutions led to the AI of today. It is aimed at students, educators, professionals, and anyone curious about how we got from Turing's 1950 question to modern AI.

It is also deliberately experimental — a playground for building with modern web and AI tooling: build-time knowledge-graph derivation, offline embeddings, automated data enrichment from Wikidata, and grounded AI question-answering. Content is curated from public sources and reviewed like code, but this is an independent side project: expect occasional gaps or inaccuracies, check the Wikipedia links on each item, and treat it as a map of the territory rather than the territory itself.

How it's built

• •Next.js 16 + React 19 + TypeScript — statically exported — every page is pre-rendered HTML, no server runtime
• •Tailwind CSS + Framer Motion — styling, dark mode, and motion
• •force-graph (WebGL) — the interactive knowledge graph
• •Fuse.js — client-side fuzzy search
• •Leaflet + Recharts — the map and chart visualizations
• •MiniLM embeddings + UMAP — the semantic "constellation" layout and "More like this" suggestions, computed offline
• •Wikidata (CC0) — enriches the knowledge graph with verified relations, refreshed monthly by an automated pull request
• •Claude (Anthropic API) — powers "Ask AI History" — answers grounded strictly in this site's own dataset
• •Azure Static Web Apps + GitHub Actions — hosting and continuous deployment

AI transparency

In line with the transparency obligations of the EU AI Act (Article 50), here is exactly where and how this site uses AI:

• •"Ask AI History" is an AI system. Answers on the graph page are generated by an AI model (Anthropic's Claude), not a human — the panel says so before you ask. Answers are constrained to this site's own dataset, carry citations, and may contain mistakes.
• •AI-generated text is machine-readable-marked. Every AI-generated answer is marked in the page markup (data-ai-generated), and the API response carries an ai_generatedfield, the generating model's name, and an X-AI-Generated header.
• •Timeline content is human-reviewed editorial content. Entries were researched and drafted with AI assistance from public sources, then reviewed and published under human editorial control; each item links to its Wikipedia article for verification.
• •Embedded videos are AI-assisted, human-selected. The YouTube videos on concept and people cards are found by an automated search-and-ranking pipeline, but every video is reviewed and chosen by a human before it appears on the site. Videos are embedded privacy-friendly (youtube-nocookie.com) and load only after you press play.
• •Nothing else. The site uses no profiling, no emotion recognition, no biometrics, no tracking of any kind — it doesn't even have analytics.

Author

Conceived and directed by Dan Paraschiv, who shaped the project's concept, scope, and direction — with substantial help from AI tooling for research, content curation, and engineering (fittingly, for a site about AI).