Human Verification: Complete Tutorial

Last week my friend Jake was adamant he'd figured out the pattern. "AI always uses capital letters," he said confidently. First round, he voted for my answer as AI because I'd capitalized properly. I'm human. He got zero points.

That's this mode in a nutshell—your assumptions about "how AI writes" are probably wrong, and the AI is getting scarily good at pretending to be human.

How This Mode Works

Everyone gets the same personal question. Could be "What did you do last weekend?" or "What's your favorite food?" or "Best concert you've been to?"

You type your answer. Meanwhile, AI also generates 2-3 fake answers trying to sound human. All answers get shuffled anonymously. Now everyone votes: which answers came from real people, which came from the AI impostor?

You get points for correctly identifying AI vs human. You ALSO get bonus points if you wrote an answer and people thought you were human (didn't vote you as AI).

How to Spot AI Answers (The Stuff That Actually Works)

Forget what you've heard about "AI always does X." Let me share what I've actually noticed after playing way too many rounds:

Pattern 1: They Sound Like They're Writing a Resume

Even when AI tries to be casual, there's often this weird formality lurking underneath.

Pattern 2: Generic AF

AI doesn't have memories or experiences. So when asked for specifics, it gives you... vibes? Concepts? Anything but actual details.

Pattern 3: Trying Too Hard to Be Fair

AI is trained to be diplomatic and considerate. Humans? We have opinions and we're not afraid to be a little bit wrong about them.

Red Flag #4: Similar Length

If 3 answers are all roughly the same length (±5 words), suspect AI generated them together. Humans naturally vary.

Red Flag #5: Lack of Filler Words

Humans use: "like", "kinda", "basically", "you know", "idk", "lol", "tbh"
AI rarely uses these unless specifically trained to mimic casualness (advanced mode).

Part 2: Writing Authentically Human

Now you know how to detect AI. Use that knowledge to avoid AI-like patterns in your own writing:

Strategy #1: Embrace Imperfection

• Skip capitals at the start of sentences
• Use "ur" instead of "your", "rn" instead of "right now"
• Add typos strategically (but not so many it looks forced)
• Forget punctuation occasionally

Strategy #2: Be Specific

Generic = AI. Specific = human. Add one unique detail:

Strategy #3: Express Emotion

Humans aren't neutral. Show feelings:

• "honestly kinda boring weekend"
• "omg that movie was so bad i walked out"
• "best day ever ngl"

Strategy #4: Use Conversational Rhythm

Write like you're texting a friend, not writing an essay:

Strategy #5: Match the Group Vibe

Read how others in your lobby wrote in previous rounds. Match their casualness level:

• If everyone's using "lol" and emojis, you should too
• If the group is more formal, adapt slightly (but stay human)
• Consistency across rounds can backfire—vary your style slightly

Advanced Techniques

The "Confidence Trap"

When voting, beware of overconfidence. If an answer looks OBVIOUSLY AI, it might be a human deliberately writing formally to trick you. Think one level deeper.

The "Timing Tell"

If someone submits instantly (under 3 seconds), they either pre-wrote it or it's AI. Genuine answers take 5-15 seconds to write.

The "Mimic Mode" Counter-Strategy

In advanced difficulty, AI analyzes human answers from previous rounds and mimics their style. To counter this:

• Vary your writing style each round (don't be predictable)
• Use inside jokes or references only humans in your lobby would know
• Include intentional "weird" phrasing that AI wouldn't generate

Mode Difficulty Levels

Lobby hosts can set AI difficulty:

Common Mistakes

Practice Exercise

Identify which of these answers is AI: