This is a submission for the June Solstice Game Jam
What I Built
Turing's Mirror is a browser game about the one question Alan Turing asked in 1950 that we still haven't answered: can you tell the difference between a human and a machine?
You read 10 conversations. For each message, you decide: human or AI?
Sounds simple. Round 1, it is. By Round 5, you'll second-guess everything.
That's the point.
๐ฎ Play it here
๐ป GitHub
Video Demo
Why Alan Turing, Why Now
June is Turing's birth month. This jam runs through June 21st โ the solstice, the longest day, the turning point of the year. The timing felt too right to ignore.
Turing gave computing its soul. Not the circuits, not the transistors โ the question. In his 1950 paper Computing Machinery and Intelligence, he didn't ask "can machines think?" He replaced that impossible question with a practical one: can a machine fool a human into thinking it's human?
He called it the Imitation Game.
We've been playing it ever since โ except now, in 2026, we're the ones being fooled. Regularly. By machines that learned from us.
I wanted to make people feel that shift. Not read about it. Feel it.
The Game Design โ Light, Dark, and the Solstice
The visual design maps directly onto the jam themes.
Dark = the machine's world. The game opens on a near-black background. Cold, clean, perfect. That's what early AI sounds like: polished in the wrong way, like someone who memorised manners without understanding them.
Light = human truth. As rounds progress and the AI gets more convincingly human, the screen warms. Win โ correctly spot the human more than half the time โ and the screen floods with light. Solstice light.
Lose, and you stay in the dark. The machines have crossed the line Turing drew.
Each round is labelled with a real year from AI history:
| Round | Year | What it represents |
|---|---|---|
| 1 | 1926 | Before Turing โ AI is science fiction |
| 2 | 1936 | Turing's first paper on computation |
| 3 | 1950 | The Turing Test is published |
| 4 | 1966 | ELIZA, the first chatbot |
| 5 | 2026 | Today: AI that passes the test regularly |
By 2026, you're trying to find humanity in text generated in milliseconds. Welcome to the present.
Code
How I Built It
No frameworks. No API keys. No backend. Just HTML, CSS, and vanilla JavaScript โ because the simplest formulation of a problem is usually the most revealing one.
The game data is an array of message pairs โ one human, one AI โ shuffled randomly so you can't pattern-match by position:
const rounds = [
{
messages: [
{ text: "Hello! I am doing well today. How may I assist you?", isAI: true },
{ text: "ugh it's so hot today lol, can't even think straight", isAI: false }
],
difficulty: "1926 โ Before the Test"
},
// rounds 2-5 get progressively harder
];
The hardest part wasn't the code โ it was writing the messages. I spent real time rewriting the Round 5 pair. Write an AI response, decide it's too obviously robotic. Make it more human. Decide it's too human now. Realize: that uncertainty is exactly the problem Turing described.
The final Round 5 pair:
- AI: "I think the complicated feeling is the honest one. The clean versions of things usually aren't real."
- Human: "that's... actually kind of what I needed to hear. weird how the right words help"
I got this one wrong myself, re-reading it cold a day later.
The ending closes on Turing's real story, in four lines:
Alan Turing (1912โ1954). He gave us the test to define intelligence. He was persecuted for who he loved. He was pardoned in 2013 โ 59 years too late. The light he switched on for computing never went out.
June is also Pride Month. Turing was gay, and the British government chemically castrated him for it. He died at 41. We live in the world he built; the dignity he was owed never arrived in time. I didn't want to make a game that skips that.
What I Learned Building This
The Turing Test is not really about AI. It's about us. The game teaches you to look for imperfection and emotional messiness as signals of humanity. The scariest moment in playtesting was when someone pointed at a human message and said "that sounds so AI." We've started writing like the models we trained.
Simplicity is a design choice. The entire game is one HTML file and one JS file. No build step. I tried to honor Turing's own instinct: reduce the problem to its essentials.
Prize Category
Best Ode to Alan Turing โ this submission honors Turing through:
- Mechanics: the Turing Test isn't referenced, it is the gameplay
- Historical narrative: each round anchors to a real year in AI history
- Visual design: light and dark as human truth vs. machine mimicry, tied directly to the solstice
- Emotional close: the full weight of Turing's contribution and the injustice he faced, during his birth month, during Pride Month
Try It
๐ฎ Play Turing's Mirror
๐ป View the code
It takes about 5 minutes to play. The Round 5 messages will make you pause.
Built solo by Tejas Patil (@tejas164321) for the DEV June Solstice Game Jam, June 2026.
Tech: Vanilla JS ยท HTML5 ยท CSS3 ยท no dependencies
Top comments (16)
The detail that stuck with me wasn't the game, it was your playtester pointing at a real human line and saying it sounded like AI. That's the actual twist, isn't it. The test was built to catch machines imitating us, and now we've drifted toward writing like them, so it kind of runs backwards. Did you notice any specific tells people leaned on early that turned out useless by Round 5? My guess is "too tidy" grammar and clean punctuation stopped meaning anything fast.
Yeah, exactly that -"too tidy" was the first tell to die. People started by guessing "fewer typos = AI" but my Round 5 AI line was full of fragments and lowercase, so that flipped completely.
The one that lasted longest: looking for hedging. Real humans hedge mid-thought ("idk it's complicated"), AI tends to land cleanly on a take even when it sounds casual. By Round 5 that gap mostly closed too, but it broke last.
Fantastic project! Turning the Turing Test into a game is a brilliant idea. It's engaging, educational, and makes players think deeply about Al and human intelligence. The design and concept are both well executed. Great work, and best of luck with future enhancements!
Nice work
Great post โฅ๏ธ
Thank You , Play the Game on provided link and visite repo on Github (Tejas164321) and star repos .......
If Guys has Any Suggesion Please Comment.......
Great work tejas ๐
Great reading!!!
the playtesting observation is the real twist โ 'too tidy' grammar dying as a tell tracks with what we see in content quality evals. automated freshness signals now get fooled by outputs that picked up human hedging patterns, not clean polished ones. the clean ones get flagged.
the round 5 pair is doing something subtle: the AI line leads with emotional resolution in a way real humans almost never do in text mid thought. that pattern survives longer than grammar as a tell.
curious if you tried this on people who write content professionally. does pattern recognition from writing work shift the round 5 results?
I never read about Alan Turing but I feel like people like to throw his name so much it's kinda becoming overused and overrated.
I'm sure the turing machine was a breakthrough in science but I feel these questions and simulations for deciding weather we're talking to an AI or Human are really overrated.
It's just like the end of Ex Machina (movie) that AI would possibly live just in the same manner. Duh, AI will be able to mimic that as humans like to make things in their own image. They will train it to be like that.
"Turning" just sounds cool so people like to throw that word around.
Round 5 hits because itโs already happening in real chats. the line between โwritten by AIโ and โwritten like AIโ is basically gone.