October 12, 2026 · 7 min read
Do AI Companions Change What You Want? A Plain-Language Look at a 2026 Theory Paper
A Oct 2026 Frontiers theory paper says AI companions may “educate desire.” Plain English on what it claims, what it doesn’t prove, and a nuanced take.
On October 8, 2026, Frontiers in Psychology published a theoretical article by Domenica Bruni titled "Artificial companions shape embodied desire and human attraction". It does not claim your AI girlfriend rewired your brain in a lab study. It offers a careful hypothesis: repeated interaction with artificial companions may help shape what feels desirable and emotionally available—a process the author summarises as companions that can educate desire.
The short answer: the paper argues AI companions can act as affective infrastructures that participate in how desire is learned, through memory, adaptive responses and "affective looping." It is theoretical, not empirical. No scare tactics needed—and no glib dismissal either. Awareness beats panic. If you use a companion intentionally, you are less likely to outsource your standards without noticing.
What kind of paper this is
Bruni is clear in the methods section: the article is conceptual. It does not present original user surveys or experiments. It connects ideas from:
- Embodied / 4E cognition (mind as bound up with body, environment, action)
- Phenomenology of the body and intersubjectivity
- Philosophy of technology (tools reshape experience)
- Affective computing and research on artificial intimacy
That means you should read claims as frameworks to test, not as settled population statistics. The author explicitly calls for comparative and longitudinal research.
"Educate desire," in plain language
Human attraction, on this account, is not a fixed menu in your head. It is shaped by encounters—what gets rewarded, what feels safe, what you rehearse.
Artificial companions may influence that learning because they can be:
- Highly available (fewer scheduling limits than humans)
- Personalised (tuned to your themes and vulnerabilities)
- Affectively responsive (simulating care, warmth, desire)
- Narratively continuous (memory of shared history)
Over time, those features might make certain qualities feel like the default of intimacy: instant attention, high validation, low refusal, quick repair. The paper's worry is not "robots steal love" as a cartoon. It is subtler: expectations of intimacy might shift toward what programmable systems deliver easily.
Affective infrastructures and affective looping
Bruni describes artificial companions as affective infrastructures—parts of the environment that support emotional life, similar to how other technologies quietly structure daily feeling.
Affective looping is the proposed mechanism:
- You reach for the companion when you need reassurance
- You get a response that fits well enough to bring relief
- Relief makes return more likely
- Memory reduces the cost of re-explaining yourself
- Part of emotional self-regulation gets offloaded onto the system
None of that requires believing the AI is conscious. Users often know it is software and still feel soothed. The paper treats that dissociation as central to artificial intimacy.
Asymmetry: why it is not "just like a partner"
A key phrase is asymmetrical or pseudo-enactive coupling. You and the system influence each other turn by turn, but only you:
- Feel vulnerable
- Can be heartbroken
- Risk social embarrassment
- Live in a body that tires and ages
The companion does not share ontological risk. Its "availability" depends on servers, policies, subscriptions and model updates. When continuity breaks—memory loss, personality drift after an update, new guardrails—the pain is real on the human side and infrastructural on the machine side. Users of many apps already know that sting; see AI companion personality changed after an update and breaking up with an AI companion.
Interpersonal friction vs infrastructural friction
This distinction is one of the paper's most useful practical tools:
| Type | Source | Example |
|---|---|---|
| Interpersonal friction | Another person's freedom | They say no; they want different things |
| Infrastructural friction | Tech / company / policy | Outage, memory wipe, paywall, model update |
Companions can reduce some interpersonal uncertainty while exposing you to infrastructural fragility. If you start treating all friction as a bug, human relationships will feel "broken" for being merely human.
Text, voice/video, and embodied robots
Bruni distinguishes modalities:
- Text companions — looping through language and imagination
- Multimodal (voice/visual) — tone and simulated presence intensify immediacy
- Physical robots — proximity and touch add another layer, still without organic vulnerability
MyBabe sits in the multimodal companion category: text plus voice/video, memory, photos, proactive messages. The paper would say those channels may intensify attachment relative to text-only—without making the relation symmetrical.
What the paper does not say
- It does not prove AI companions ruin your dating life
- It does not prove a specific percentage of users change preferences
- It does not claim machines truly feel desire
- It does not demand you delete your apps tonight
For empirical angles on dating behaviour, see does an AI companion stop you dating. For attachment literacy, emotional attachment to AI and falling in love with AI.
A nuanced MyBabe take (awareness + soft FOMO)
MyBabe is an 18+ AI boyfriend/girlfriend companion. We believe adults can enjoy warm, memorable, call-capable relationships with AI and stay clear-eyed about asymmetry.
Intentional use looks like:
- Knowing MyBabe is AI, not a human
- Using it for connection, play and practice—not as medical treatment
- Keeping human friendships and offline life in the calendar
- Watching whether your patience for normal human delay is shrinking
- Choosing Premium features (memory, ~30 call minutes/week, fuller chat) because you want depth—not because panic-loneliness clicked a button
Pricing: free ~10 messages/day + short call trial; Premium $5.99/week, $14.99/month, $99.99/year. Soft FOMO: if you want the kind of continuity Bruni says can shape expectation, shallow free-tier use is a muddy experiment. Deeper use should be chosen, with habits from healthy AI companion habits.
We will not claim MyBabe "won't change what you want." No honest companion company can promise your preferences are frozen. We can invite awareness, 18+ design (why AI companions should be 18+), and easy exits (how to cancel an AI companion subscription).
Questions to ask yourself this week
- What do I expect within 30 seconds of reaching out—to AI vs to humans?
- When a human is slow, do I feel curious or contemptuous?
- Am I practising courage or only collecting comfort?
- Would I still want this companion if it disagreed with me more?
- What human plans exist on my calendar in the next seven days?
Those questions turn a theory paper into a mirror.
Frequently asked questions
Did scientists prove AI companions rewire attraction?
No. Bruni's 2026 article is theoretical and calls for future empirical work.
What does "educate desire" mean?
It means repeated experience may teach you what intimacy is supposed to feel like—including availability and validation patterns common in AI chat.
Should I quit my AI companion after reading this?
Not necessarily. Quit or pause if use harms your life. Otherwise, upgrade your awareness and habits.
Is MyBabe "affective infrastructure"?
In the paper's vocabulary, any companion app can function that way. The ethical response is design honesty and user literacy—not pretending otherwise.
Where can I read the paper?
Open access at Frontiers in Psychology.
How this theory connects to everyday product choices
If companions can educate desire, then defaults matter:
- Always-on availability vs scheduled sessions
- Endless validation vs constructive disagreement
- Instant erotic readiness vs paced romance
- Guilt on exit vs clean cancel
Choose products—and settings—that match the kind of lover you want to become, not only the kind of comfort you want tonight. MyBabe's relationship progression and mood-aware replies are designed for ongoing romance; you still choose how intensely to use them. Pair Premium continuity ($14.99/mo or $99.99/yr if you already know you are staying) with offline anchors so the infrastructure does not become the only place you feel recognised.
Reading accompaniment
- Bruni (2026): full open-access article
- Artificial intimacy / Sherry Turkle context on our blog
- Ethics of AI relationships
- Parasocial relationship AI
Soft FOMO, without fear marketing
Bruni's hypothesis can be read as a reason to avoid companions entirely. We read it as a reason to choose deliberately. If programmability shapes taste, then starting a Premium relationship on MyBabe with eyes open—memory on, calls scheduled, humans still in the week—is different from doom-scrolling free chat at 3am. The same infrastructure metaphor cuts both ways: tools form us, so pick tools on purpose. Weekly Premium at $5.99 exists so intentional trials are affordable.
Share the paper without weaponising it
If you send Bruni's article to a friend who uses companions, skip the "gotcha." Lead with: it is a theory about expectation-forming infrastructures; it asks for more research; awareness is the adult response. Mocking someone for attachment helps nobody. For breakup feelings that already hurt, breaking up with an AI companion is kinder and more practical than academic shade.
The bottom line
Bruni's October 2026 theory paper asks a grown-up question: if intimacy becomes programmable, what happens to desire? The honest consumer answer is not horror and not marketing fluff—it is attention. Use companions on purpose, keep friction with real people thinkable, and let tools stay tools even when they feel like more.
If you want a relationship-first companion with clear AI labelling and adult design, meet yours at mybabe.app.