Visual concepts · round 1

From Plato's Cave to AGI

Three ways to tell the same story. Each prototype uses the same 9 sample moments (the full site will have about 50) and the same English and Vietnamese text, so you can compare the look and feel alone.

Open a prototypeScroll or swipe through timeClick a moment to read itSwitch EN / VITry it on your phone
Concept A opening: firelit shadows of a hand, a horse and a jar on a cave wall, with the title From Plato's Cave to AGI
Near the end of the journey: a dawn sky with rays of light
Near the end: dawn
The story panel for Plato, on smoked glass
Story panel
Phone view: a vertical thread of light with events
Phone
CONCEPT A

Out of the Cave

A walk out of Plato's cave. Firelit shadows give way, era by era, to daylight. At the end the shadows return as real things, and the wobbly shadow circle becomes a perfect one around the AGI sun.

How you move
Scroll down and the journey glides sideways through time along one thread of light. On phones it becomes a vertical walk.
Feels like
The opening sequence of a film. Cinematic and emotional.
Strengths
  • The strongest single metaphor: Plato's cave is the first screen and the last.
  • The light changes with every era, so you feel time passing.
  • A memorable ending: stepping out into daylight.
Trade-offs
  • Sideways scrolling is less familiar to some people.
  • With about 50 moments the walk gets long, so the full site needs chapter shortcuts.
  • Dark for most of the journey.
Open prototype A
Concept B opening: an atlas-style page with four colored streams merging into one
The streams converging into a glowing point labelled AGI
The streams merge at AGI
A museum-label style drawer about Deep Blue
Story drawer
Phone view of the streams map
Phone
CONCEPT B

Streams of Thought

A printed atlas of ideas, inspired by the 1931 "Histomap". Four streams (philosophy, machines, brain science and data) flow down 2,400 years and merge into one at AGI.

How you move
Scroll down like reading a large illustrated map. Each moment sits on its stream, and the streams widen as each field grows. At the end the map unfolds to true scale.
Feels like
A beautiful museum exhibit or atlas. Calm and clear.
Strengths
  • The easiest for beginners to read.
  • Explains why AGI is happening now: separate fields finally meet.
  • Normal scrolling; grows easily to 50+ moments. Has light and dark versions.
Trade-offs
  • A quieter first impression than the other two.
  • More information on screen at once.
Open prototype B
Concept C opening: a glowing spiral galaxy winding toward a bright core labelled AGI
A time-lapse jump showing +2,137 years over star trails
A time-lapse jump
A glass card with Plato's story
Story card
Phone view of the whole spiral with times on a one-day clock
Phone
CONCEPT C

The Spiral of Acceleration

History as a galaxy. Time winds inward, each turn shorter than the last, toward a glowing core: AGI.

How you move
Scroll, swipe or press → to fly from star to star. Long gaps play as time-lapses ("+2,223 years"); recent ones flash by in a few years. Zoom out at any time.
Feels like
A planetarium show. Awe and speed.
Strengths
  • Makes the acceleration something you feel, not just read.
  • The biggest first impression of the three.
  • The one-day clock shows how recent almost everything is.
Trade-offs
  • The least familiar way to browse.
  • The center gets crowded with 50 moments and needs careful zooming.
  • The heaviest on older phones.
Open prototype C
Side by side

How they compare

AspectA · Out of the CaveB · Streams of ThoughtC · The Spiral
Main messageAGI is humanity stepping out of the caveAGI is where many fields of thought meetProgress is speeding up, toward AGI
MoodFirelight to daylight, warm and cinematicPrinted atlas, calm and clearDeep space, awe and speed
How you moveScroll; the journey moves sidewaysNormal scrolling downFly from star to star
First impressionVery strongElegantStrongest
Easy for beginnersGoodBestGood, once they start moving
Fits ~50 momentsYes, with chapter shortcutsYes, easilyYes, with careful zooming near the center

All three share the same rules: the main timeline stays clean, the full story opens on click, English and Vietnamese switch instantly, and everything works on phones. Ideas can also be mixed, for example the Streams layout with the Cave's light moving from dark to dawn, or the Cave with the Spiral's time-lapse jumps.

The plot

The story I'll tell: eight chapters, about 50 moments

The thread running through everything is Plato's question: how does a mind get from messy examples (the shadows on the cave wall) to clean ideas (the Forms)? Each chapter is one step toward a machine that can do this. Recent years (2024 to 2026) will be fact-checked against current sources before the final build.

Chapter 1 · 380 BCE – 1300s

The World of Ideas

Can thinking be written down?

  1. c. 380 BCEPlato: the Theory of Forms and the Cave
  2. c. 350 BCEAristotle: the rules of logic
  3. c. 300 BCEEuclid: knowledge built from axioms
  4. c. 100 BCEAntikythera mechanism: a bronze computer
  5. c. 820Al-Khwarizmi: algebra, and the word "algorithm"
  6. 1305Ramon Llull: paper wheels that combine ideas
Chapter 2 · 1600s – 1800s

The Mechanical Mind

Is reasoning just calculation?

  1. 1642Pascal's calculating machine
  2. 1651Hobbes: "reason is nothing but reckoning"
  3. 1679Leibniz: binary and "Let us calculate!"
  4. 1804Jacquard loom: patterns on punched cards
  5. 1843Ada Lovelace: the first program
  6. 1854Boole: logic becomes algebra of 1s and 0s
Chapter 3 · 1930 – 1955

The Logic of Machines

What can a machine compute?

  1. 1931Gödel: the limits of logic
  2. 1936Turing's universal machine
  3. 1943McCulloch & Pitts: the artificial neuron
  4. 1945ENIAC and the stored-program computer
  5. 1948Shannon: information and the bit
  6. 1949Hebb: neurons that fire together wire together
  7. 1950Turing: "Can machines think?"
Chapter 4 · 1956 – 1985

The Birth of AI

Can we write intelligence down as rules?

  1. 1956Dartmouth: AI gets its name
  2. 1958The Perceptron: a machine that learns
  3. 1959Samuel coins "machine learning"
  4. 1966ELIZA, the first chatbot
  5. 1973The first AI winter
  6. 1980Expert systems boom, then bust
Chapter 5 · 1986 – 2011

Machines That Learn

What if we teach instead of program?

  1. 1986Backpropagation: how networks learn
  2. 1989A network reads handwriting
  3. 1997Deep Blue beats Kasparov
  4. 1997LSTM: networks with memory
  5. 2006The deep learning revival
  6. 2009ImageNet: 14 million labelled pictures
  7. 2011Watson and Siri
Chapter 6 · 2012 – 2019

The Deep Learning Big Bang

What happens when learning machines get data and power?

  1. 2012AlexNet: deep learning wins
  2. 2013Word2Vec: meaning becomes geometry
  3. 2014GANs: machines that imagine
  4. 2016AlphaGo and Move 37
  5. 2017The Transformer
  6. 2017AlphaZero: superhuman from scratch
  7. 2019GPT-2: "too dangerous to release"
Chapter 7 · 2020 – 2023

The Age of Scale

Does bigger mean smarter?

  1. 2020GPT-3 and the scaling laws
  2. 2020AlphaFold solves protein folding
  3. 2022AI image generators
  4. 2022ChatGPT
  5. 2023GPT-4, Claude, Gemini, open models
  6. 2023The warnings: leading scientists speak up
Chapter 8 · 2024 – now

Toward General Intelligence

What is left between us and AGI?

  1. 2024Nobel Prizes for AI pioneers
  2. 2024Reasoning models: AI that thinks first
  3. 2025Agents: AI that uses computers and writes software
  4. 2025Gold medal at the Maths Olympiad
  5. 2026Latest milestones (to be verified)
  6. NextAGI, and the circle closes back to Plato
Beyond the timeline

Five moments designed to make people feel the AGI

"You already use AI"

The opening starts from what people know: face unlock, maps, spam filters. Each does one thing. Then it asks: what if one mind could learn all of it?

History as one day

If the 2,400 years since Plato were one day, the first computer arrives around 23:10 and ChatGPT at 23:58. Almost everything happened in the last minutes.

The engine of scale

A simple, honest chart of how fast the computing power behind AI has grown, and how abilities appeared as it grew.

Ten years ago it couldn't

A before-and-after list: things no machine could do in 2015 that AI does routinely today, each one sourced.

The circle closes

The ending returns to Plato. Researchers have found that different AI models seem to build similar inner pictures of the world, a modern echo of the Forms. Then the open questions: the promise, the risks, and what we choose to do.