Proof of possibility, kept in one place

AI Achievement
Atlas.

From foundational papers to your team’s smallest hard-won breakthrough. One place to collect what AI made possible.

Explore the atlas
Museum display of a silicon wafer, circuit artifacts, molecular model, and Go stones
Starter collection
plus your archiveA

The record,
still unfolding.

A curated starter set of widely documented milestones, ready to become your own archive. Search the collection or add the achievement you want remembered.

6 shown

Progress is not a straight line.

It arrives as a move no machine could make, a structure no lab could see, or a method that changes every system after it.

1956

A field finds its name.

The Dartmouth workshop proposal gave “artificial intelligence” a shared banner and a research agenda ambitious enough to organize generations of work.

2012

Vision changes course.

AlexNet’s ImageNet performance helped make deep convolutional networks the new center of gravity for computer vision.

2020

Biology gains a new instrument.

AlphaFold2 demonstrated striking protein structure prediction accuracy, turning a long-standing scientific challenge into a practical computational capability.

“An achievement matters twice: when it happens, and when someone keeps the record.”

Make the atlas yours.

Filtering, ordering and local storage in the achievement atlas

Read the explanation

The atlas begins with six supplied starter records. Three have Research category, two Systems and one Science. Choosing Research restricts that set to three. Searching twenty seventeen within Research keeps one because the filter joins title, description, year and category into a case insensitive substring test. Category and query conditions combine with logical and. The count describes visible records, not total stored records or a search of external research. A custom addition appends one object to local browser storage. Six starter records plus one custom record gives seven. Rendering merges both arrays and sorts by descending numeric year. Consequently an older new record appears lower in the timeline rather than automatically at the top. Adding resets the query and category to show the full set. The identifier uses the current timestamp, and the image URL is derived from title; neither establishes independent evidence that the achievement happened. Export calls the unfiltered combined list and removes only each record image field, so a Research filter showing three starters still exports all six starter records. Evidence links accept only HTTP or HTTPS URLs; accepting a URL does not fact check its contents. Reset requires a second click within five seconds, then removes custom additions while preserving starters. This explainer verifies local collection operations and supplied counts. It does not update historical claims or certify external evidence, and local storage is not a server backup.

Add an achievement

Achievement record

Local Collection Mechanics & Schema Architecture

How does AI Achievement Atlas manage client-side state, multi-field filtering, and timeline ordering?

AI Achievement Atlas operates entirely within browser memory by merging a six-item starter dataset with user records saved under a single localStorage key. When searching or selecting category filters, the script performs case-insensitive substring checks across title, description, year, and category fields, sorting the union in descending order by milestone year.

The application performs naive client-side substring matching without stemming, semantic indexing, or ranking. Custom records are restricted to the local browser profile and device unless manually exported via JSON; clearing browser cache or running across separate browsers does not synchronize milestones.

Try a worked example

Clicking the 'Research' category filter updates its active state and reduces the visible milestone count from 6 shown to 3 shown (displaying Attention changes the architecture, AlexNet reshapes vision, and A field gets its name). Clearing or selecting 'All' restores the full 6 records.

State Composition & Sorting Model

The atlas combines an immutable hardcoded array of six foundational achievements (Dartmouth 1956, Deep Blue 1997, AlexNet 2012, AlphaGo 2016, Transformer 2017, AlphaFold2 2020) with dynamic user items loaded from localStorage under key 'aiAchievementAtlas:v1'. The combined list is re-sorted chronologically descending by numerical year whenever a render cycle executes.

Filtering & Query Evaluation

Filtering evaluates both active category buttons and search input text. A milestone passes if its category strictly matches the selected filter (or if 'All' is active) and all entered whitespace-trimmed query tokens appear within the concatenated string of title, description, year, and category.

Data Portability & Validation Scope

User submissions validate that title and description are non-empty strings and that evidence URLs conform to standard http or https protocol formats. Export bundles metadata and entry properties into an RFC-standard JSON blob without sending information to an external server.

Super generates helpful tools and automates fact-checking across the internet proactively. If you enjoyed this tool, build your own with Super and share it with a friend.