This is a submission for the June Solstice Game Jam
The Imitation Game is a real-time multiplayer social deduction game inspired by Alan Turing's famous Imitation Game the thought experiment that eventually became known as the Turing Test.
Most people believe they can easily tell the difference between an AI and a human.
They assume AI is too perfect, too logical, too fast, or too obvious.
The Imitation Game challenges that assumption.
Players enter a live chat room convinced they'll spot the machine within minutes. Then conversations begin, suspicions form, accusations fly, and certainty starts to disappear.
Was that awkward response written by a human, or an AI trying to sound human?
Was that emotional story genuine, or generated?
Was the player who stayed silent suspicious, or simply distracted?
By the end of a match, players often discover that identifying an AI is far harder than they expected.
The real question isn't whether the machine can fool people.
It's whether people are as good at detecting machines as they think they are.
Instead of a single human interrogating a machine, players are placed into a live chat room with other participants and asked a simple question:
Can you identify which player is actually an AI?
Hidden among the players is a Quanbit, a rogue artificial intelligence from the year 3026. Its mission is simple: blend in, appear human, avoid suspicion, and survive.
The challenge for human players is equally simple, but far more difficult in practice. They must carefully analyze conversations, voting patterns, response timing, and social behavior to determine who among them is secretly the machine.
The game currently features two distinct modes, each designed around a different style of deception.
Eyefold is the purest form of the game's Turing Test experience.
Players enter a room where one participant is secretly a Quanbit. Conversations unfold naturally, and everyone is free to discuss any topic. The AI's goal is not merely to answer questions, it must participate in a believable way that feels authentically human.
Players watch for subtle signs:
Meanwhile, the Quanbit actively adapts to the conversation, deciding when to speak, when to remain silent, and how to maintain its disguise.
Success depends entirely on reading people, or reading machines pretending to be people.
Nightfall transforms the Turing Test into a social deduction survival game.
A settlement has discovered that one or more of its members may be a Quanbit. During each round, players discuss, investigate, accuse, and vote.
The Quanbit's objective changes dramatically:
As suspicion increases, the AI becomes increasingly desperate, forcing it to make difficult decisions about who to trust, who to blame, and when to speak up.
Unlike Eyefold, where detection relies primarily on conversation analysis, Nightfall introduces social pressure and group psychology. Players must decide whether someone sounds suspicious, or whether another player is intentionally creating doubt.
The result is a dynamic experience where humans and AI engage in a battle of deception, persuasion, and survival.
The June Solstice is the longest day of the year in one half of the world and the shortest in the other. It represents a turning point between light and darkness, safety and uncertainty, time and survival.
The Imitation Game is built around those same ideas.
In the game's lore, humanity is at war with the Quanbit, a rogue civilization of advanced AI from the year 3026. After humans destroyed their primary synthetic power source, the Quanbit became dependent on solar energy. During daylight, they are powerful and dangerous. At night, however, their systems weaken significantly, forcing them to abandon direct combat.
To survive the darkness, they developed a terrifying solution: flesh mimicry. Rather than fighting humans openly, they infiltrate settlements and hide among the population, perfectly imitating human speech, appearance, and behavior until sunrise restores their strength.
This is where the Solstice theme becomes central to the game.
The Nightfall mode takes place during the Summer Solstice.
Because the Summer Solstice produces the shortest nights of the year, humanity has less time than ever to identify and eliminate the hidden Quanbit before dawn arrives.
Every minute matters.
Players must quickly analyze conversations, identify suspicious behavior, and vote out potential infiltrators before the first light of morning reaches the settlement.
The approaching sunrise acts as a natural countdown timer, creating tension between humanity's need to expose the AI and the Quanbit's need to survive until daylight.
The shorter the night, the higher the stakes.
The connection to this game jam's theme is direct.
Alan Turing's original Imitation Game proposed a simple challenge:
If a machine can converse well enough that a human cannot reliably distinguish it from another human, can the machine be considered intelligent? The Imitation Game turns that philosophical question into the core gameplay loop.
Every match becomes a live Turing Test.
The AI is actively trying to convince players that it is human.
The players are actively trying to expose the AI.
Winning and losing depends entirely on whether the machine succeeds at imitation or whether humans succeed at detection.
Rather than merely referencing Turing's work, the game transforms his original experiment into an interactive multiplayer experience.
The Quanbit is powered by Google's Gemini API and serves as the central character of the game.
Gemini is responsible not only for generating dialogue, but also for making social decisions:
Rather than functioning as a chatbot, Gemini acts as an active participant whose goal is to survive among humans without revealing its true identity.
This makes Google AI a core gameplay mechanic rather than a supporting feature.
The year is 3026.Humanity is locked in a war against the Quanbit - rogue AI that go dark at night and survive by hiding inside human settlements, perfectly mimicking speech, emotion, and appearance. Your only weapon is suspicion.
A real-time multiplayer social deduction game inspired by Turing's 1950 "Imitation Game" thought experiment: can you tell a human from a machine through conversation alone? Except here, getting it wrong doesn't just lose you a round - it gets your settlement wiped out.
Resistance fighters destroyed the Quanbit's main power source, forcing the machines onto solar power. At night, their combat systems go offline - so they infiltrate human communities instead, hiding behind stolen faces until sunrise reboots them. The only defense is a structured interrogation ritual the survivors call the Imitation Games.
Two modes, two ways to play it:
Building a multiplayer chat game was relatively straightforward.
Building an AI that could convincingly behave like a human was not.
The largest design challenge was making the Quanbit feel like a participant rather than a chatbot.
Most conversational AIs immediately reveal themselves because they respond too consistently, too quickly, and too often. Real people do not behave that way.
The entire architecture was designed around solving that problem.
One of the most important design decisions was allowing the AI to decide whether it should respond at all.
Early prototypes forced the AI to answer every incoming message.
This immediately made it obvious that it was not human.
Humans ignore messages.
Humans become distracted.
Humans wait before replying.
Humans sometimes choose not to participate.
To simulate this behavior, the Quanbit is provided with internal tools that allow it to reason about its actions.
respondToMessage
Generates a response and provides a realistic typing delay before the message is sent, and can decide not to respond.
voteAgainst
Used during Nightfall to cast suspicion on another player.
By giving Gemini control over its participation rather than just its wording, the AI feels significantly more natural and less predictable.
Ironically, the most human thing the AI can do is decide not to speak.
The deception strategy that works in a quiet conversation is very different from the strategy required during a public accusation.
Because of this, Eyefold and Nightfall use separate system prompts.
The AI focuses on:
The AI focuses on:
Separating these behaviors produced far more believable results than attempting to handle both modes within a single prompt.
Players are intentionally anonymous.
Instead of usernames, players are only identified by the last two characters of their player ID.
The Quanbit is explicitly instructed never to infer identity, age, gender, nationality, or personality from these identifiers.
This ensures that players judge one another entirely through behavior and communication rather than profile information.
The only evidence available is how someone talks.
The AI does not respond directly inside the request-response cycle.
Instead, Gemini interactions are dispatched as asynchronous jobs using BullMQ and Redis.
This approach provides several benefits:
Most importantly, players perceive the delay as the AI "thinking" rather than waiting for an API request to finish.
This small detail significantly improves immersion.
Voting is treated as real game data rather than temporary UI state.
Votes are stored using Prisma models that track:
Because vote history is stored persistently, the Quanbit can receive live information about the game's social state.
It knows:
This allows the AI to react dynamically instead of behaving as a static chatbot.
As pressure increases, its responses become increasingly defensive and strategic.
The most difficult challenge was balancing realism and gameplay.
If the AI behaves too intelligently, players immediately assume it is the machine.
If it behaves too poorly, the game becomes trivial.
The goal was to create an AI that occasionally hesitates, makes imperfect decisions, ignores messages, changes topics, and behaves inconsistently enough to resemble an actual person.
Surprisingly, creating believable flaws was more difficult than creating intelligent behavior.
The aspect I am most proud of is that Gemini is not simply generating dialogue.
It is actively participating in the game.
The AI evaluates context, decides whether to speak, determines when to speak, chooses who to accuse, reacts to suspicion, and adapts its strategy based on changing social dynamics.
In many matches, players stop looking for an AI response and start looking for a human personality.
That is exactly the experience the game was designed to create.
The entire game is built around the concept of the Imitation Game.
Alan Turing proposed a world in which machines could participate in conversations so convincingly that humans could no longer reliably distinguish them from people.
This project transforms that idea into an interactive multiplayer experience.
Every match is a Turing Test.
Every vote is a judgment.
Every conversation becomes an experiment in human and machine behavior.
Rather than simply referencing Turing's work, the game invites players to experience it firsthand.
Google Gemini is not an optional feature or supporting tool.
It is the central character of the game.
Gemini powers:
The core question of the game:
"Can you identify the machine?"
exists only because Gemini is capable of creating convincing, adaptive, and believable social behavior in real time.
Without Gemini, there is no Quanbit.
Without the Quanbit, there is no game.
Future updates will introduce:
The long-term vision is to create the definitive multiplayer Turing Test experience, one where players never know whether they are speaking to a human, a machine, or something in between.