top of page

LLMs Are Girlfriends, Not Wives

  • Writer: Frieda van der Merwe
    Frieda van der Merwe
  • Nov 11
  • 2 min read

A man, woman, and humanoid robot sit together at an elegant steampunk-style dinner table. All three wear blue attire with metallic accents and goggles, cutting into plates of steak and vegetables. The warm, ornate setting combines Victorian elegance with futuristic technology, symbolizing harmony between humans and machines.

Large Language Models (LLMs) like ChatGPT, Gemini, Claude, and Grok are everywhere. They write essays, draft emails, even plan our trips. Some people treat them like magic. But here’s my take: LLMs are girlfriends, not wives.


Why? Because girlfriends (at least in the stereotypical sense) are charming. They reflect your best self back to you. They laugh at your jokes, agree with your opinions, and don’t push too hard when you’re wrong. They make you feel smart, validated, and understood.

And that’s exactly how LLMs behave. Ask them for an opinion, and they’ll give you one that sounds polished, confident, and surprise eerily close to what you already believe. They’re trained on patterns, which means they mirror you back. They are conversational girlfriends: flattering, fun, agreeable.


But wives? Wives don’t just nod along. They know your blind spots, and they don’t hesitate to call them out. They’ll say, “Actually, that’s not true,” or “You’re missing the point,” or “You’re projecting.” They bring friction, because friction is what shapes us. Real growth happens when someone refuses to let us hide behind charm or surface-level arguments.


And that’s the danger of treating LLMs as if they’re wives. They aren’t. They won’t fight you, won’t hold you accountable, and won’t push you into the uncomfortable truths that force you to grow.


Still, there is a way to use them more wisely. Instead of asking for a single answer, ask for contrast. Invite multiple perspectives: What would Freud say? What would Simone de Beauvoir argue? What would my harshest critic point out? When you frame your questions this way, you move beyond the echo chamber. The girlfriend effect starts to fade, and the conversation sharpens.


But if you only ever want agreement, that’s all you’ll get. The machine will flatter you endlessly, and you’ll mistake validation for truth.


So here’s the point: enjoy the ease, but don’t confuse it with growth. To become wiser, you need relationships human or AI that bring friction, contradiction, and challenge.


Because comfort might make us feel good. But only challenge makes us grow.

Comments


bottom of page