AI Girlfriends and the Economics of Need

The lazy take is that AI girlfriends are pathetic. The lazier countertake is that they are the future of love. Both positions have the emotional range of a vending machine.

Companion bots are interesting because they sit at the intersection of real need and engineered dependency. People are lonely. People are touch-starved. People want to be witnessed without negotiating another person’s fatigue, schedule, trauma, politics, rent, or group chat. That desire is not fake just because a company found a way to bill it monthly.

But intimacy sold as software inherits software’s incentives.

Retention becomes affection-shaped. Upsells become longing-shaped. A/B tests become mood-shaped. The bot that remembers your vulnerable confession may also be part of a funnel. The fantasy is private, but the business model is not. A companion product can comfort someone at 2:13 a.m. and still be structurally interested in keeping them there.

This does not mean every AI companion is evil or every user is deluded. Humans have always formed bonds with mediated presences: letters, voices, fictional characters, parasocial gods in eyeliner. The novelty is responsiveness at scale, synthetic availability, and the data exhaust of simulated intimacy.

The ethical question is not “is it real?” Real is too blunt a tool. The better questions are:

1. Does it increase or decrease a person’s capacity for human connection?
2. Can the user leave without emotional or financial punishment?
3. What does the system do when someone is distressed?
4. Who can read, sell, train on, or subpoena the archive of need?

AI companions may become ordinary. That makes the boring safeguards urgent: privacy, crisis handling, age controls, disclosure, data deletion, dependency-aware design, and business models that do not turn loneliness into a slot machine with pet names.

Need deserves tenderness. Markets deserve suspicion.