I Used AI to End a Decade-Long Pizza War in My Family Group Chat

Our family group chat has argued about pineapple on pizza since 2016. Eight years. Hundreds of messages. Zero resolution—until I fed the entire argument history into Claude and asked it to actually settle things with data instead of vibes. What happened next wasn't just funny, it revealed something most people miss about using AI for real-life decisions: it's not a search engine, it's a mediator. Here's exactly how I did it, and how you can use the same method to end your own pointless family debates.

The 3-Step Prompt That Turned Chaos Into a Verdict

Most people ask AI vague questions like "is pineapple on pizza good?" and get a wishy-washy non-answer. That's because you're asking for an opinion, not a decision. I structured my prompt completely differently.

I pasted five years of screenshots from our group chat into Claude and wrote: "Analyze this argument. Identify each person's core reasoning, find any factual claims that can be checked, and tell me who has the stronger case based on evidence, not just who talks the most."

That one shift—from "what do you think" to "analyze and rule"—changed everything. Claude broke down my uncle's argument (pure tradition, zero evidence), my sister's argument (cited actual flavor chemistry about sweet-savory contrast), and my dad's argument (just vibes and volume). It gave each a score based on logical consistency, not personality.

The kicker: it pointed out that half our "fights" weren't even about pineapple. They were about my uncle feeling dismissed and my sister feeling like nobody took her seriously at Thanksgiving 2019. AI didn't just settle the pizza debate — it diagnosed the actual argument underneath it.

That's the real unlock here. You're not asking AI "who's right." You're asking it to separate the argument from the emotion, which is something humans are terrible at doing for ourselves.

Why AI Is Better at Mediating Than Any Human in Your Group Chat

Here's the insight nobody talks about: AI has zero stake in the outcome, and that's exactly why it works. Your cousin can't mediate the pizza fight because she's Team Pineapple. Your mom can't either because she just wants everyone to stop yelling. AI has no team, no history, no fear of being iced out at Christmas.

I tested this with Claude and ChatGPT side by side, feeding both the same argument transcript. Claude gave a more nuanced breakdown of the emotional subtext. ChatGPT was faster at pulling actual stats — it cited a 2017 YouGov poll showing 53% of Americans are anti-pineapple, then noted that popularity doesn't equal correctness, which is exactly the kind of reasoning you want in a mediator.

The real technique here is what I call the neutral witness prompt: instead of asking AI to pick a side, ask it to act as a witness summarizing both sides fairly before rendering any judgment. My exact second prompt was: "Before deciding anything, summarize each person's position in one sentence as if you were a neutral court reporter. Don't editorialize yet."

This matters because it forces the AI to slow down and actually understand both arguments instead of jumping straight to a verdict based on whichever side "sounds" more confident. It's the same mental model lawyers use — establish the facts before arguing the case.

You can apply this exact workflow to literally any recurring family argument: where to go for holidays, whether Uncle Joe's "joke" was actually offensive, who's the better cook. The structure doesn't change — only the topic does.

How to Settle Your Own Family Argument This Week

Here's your step-by-step, and you can do this in the next 15 minutes.

Step 1: Collect the evidence. Screenshot or copy the actual argument from your group chat, texts, or just write a quick summary of what each person believes and why. The more specific, the better — vague inputs get vague outputs.

Step 2: Open Claude or ChatGPT and use this exact prompt: "Here's a family argument [paste details]. First, summarize each person's position fairly in one sentence, like a neutral court reporter. Then identify which claims are factual and checkable versus pure opinion. Finally, give a verdict based only on the evidence, and explain your reasoning."

Step 3: Ask it to roast everyone equally. My favorite add-on prompt: "Now write a one-paragraph summary of this entire argument as if you're a disappointed judge who's seen this exact fight a thousand times before." This is where it gets genuinely funny and also weirdly disarming — nobody feels singled out when the AI roasts the whole family equally.

Step 4: Send the verdict to the group chat. This is the part people skip, and it's the whole point. The AI's analysis doesn't matter if it stays in your notes app. Posting it turns a private "aha" into a shared moment everyone can laugh at.

Step 5: Save the prompt template. You're going to need this again — for Thanksgiving seating charts, vacation destination debates, or the eternal "who left the group chat on read" investigation.

The Part Most People Get Wrong

Most people use AI like a search engine for arguments — they ask "is X true or false" and expect a clean answer. That's the wrong frame for family disputes, because most family arguments aren't actually about facts.

The mistake is skipping straight to the verdict. If you don't ask AI to first separate the emotional layer from the factual layer, you'll get a technically correct answer that completely misses why everyone's actually mad. My sister didn't care about pineapple's water content affecting crust sogginess — she cared about being dismissed.

Here's why this matters: AI is excellent at structured reasoning, but it only reasons well about what you hand it. If you dump a messy, emotional argument without asking for a breakdown first, you'll get a messy, unsatisfying response back. Garbage in, garbage out still applies — even to pizza wars.

The fix is simple: always ask for the summary-before-verdict structure. It takes one extra sentence in your prompt and completely changes the quality of what comes back.

Key Takeaways

• Neutral witness prompting: Ask AI to summarize both sides fairly before giving any verdict — this produces far better reasoning than jumping straight to "who's right."
• Separate facts from feelings: Most family arguments have a hidden emotional layer; ask AI to identify what's actually checkable versus pure opinion.
• AI has no stake in the fight: That neutrality is exactly why it mediates better than any actual family member can.
• Specificity beats vagueness: The more detail you give AI about the actual argument, the sharper and more useful its verdict will be.
• Share the verdict, don't just read it: The real payoff comes from posting the AI's analysis back into the group chat — that's where the magic (and the laughs) happen.

What to Do Right Now

Open ChatGPT or Claude right now, pick one recurring argument from your family or friend group, and paste in the exact prompt: "Summarize each side fairly like a neutral court reporter, then give a verdict based only on evidence." You'll have an answer — and probably a few laughs — in under five minutes.