The short version: "should I stay or leave" is one of the most consistently high-engagement questions on Reddit — one thread on staying too long in a relationship pulled 219 votes and 147 comments; another on knowing when to leave pulled 253. There's no factual answer to hand someone here, from an AI or anyone else. What actually helps is something narrower: getting clear on which specific factor is really driving the doubt, because that's the part a decision this personal usually depends on.
This isn't a replacement for a therapist, a counselor, or people who actually know your situation — for anything involving safety, it's not the right tool at all. What it can do is what a second opinion always does: help you see the shape of your own reasoning more clearly.
A question people ask each other constantly, and rarely resolve with one answer
The pattern across the threads is consistent: people aren't usually asking "is this relationship objectively bad" — they're asking because they already sense the answer and want to know if it's reasonable to trust that instinct. "Torn, should I stay or should I go" doesn't need someone to tell them yes or no. It needs someone to help them see what, specifically, is generating the doubt.
Why one confident answer — AI or otherwise — isn't actually useful here
A single AI asked "should I stay with my partner" will produce something that sounds like an answer, built from a description that's necessarily partial — you're the only one who has the full picture, and even you may not have named the actual deciding factor yet. Different models, given the same description, can genuinely emphasize different things: one might weight the trust-breach details more heavily, another might weight the pattern of behavior over time, a third might focus on what you said about your own hopes for the relationship. That's not models being unreliable. It's the same question producing different valid readings, because the question itself is a judgment call, not a fact to retrieve.
What actually helps: naming the real tradeoff
- The disagreement, if you ask more than one source, isn't noise — it's information about which factors are genuinely in tension. If every reading converges on the same concern, that's worth taking seriously. If they diverge, that split usually maps to the real fork in your own thinking.
- Ask the forced-choice version, not the open one. Not "should I stay or go" but "which regret would be harder to carry — leaving something that still had love in it, or staying in something you already know isn't right?" That kind of question surfaces what you actually believe faster than a general one does.
- Come back and refine. The first pass rarely settles it. Once you notice which factor the disagreement is really about, ask again with that named explicitly — the second, sharper conversation is usually more useful than the first.
Satcove runs a question like this past six AI models at once — Claude, GPT, Gemini, Mistral, Perplexity and Grok — not to hand you a verdict, but so you can see where different readings agree, where they don't, and refine from there. For anything involving safety or abuse, please reach out to a local support service or hotline instead — this is not the right tool for that.
Trying to get clarity on a decision that's really yours to make? Talk it through with 6 AIs at once.