October 11, 2026 · 7 min read
Does an AI Companion Stop You Dating? What a Randomized Experiment Found
Does using an AI girlfriend or boyfriend make you date less? A 2026 six-week randomized experiment and new interview research give a nuanced answer.
One of the most common worries about AI girlfriends and boyfriends is simple: if an AI is always there, always kind and never rejects you, will you stop bothering with real dating? Critics say yes. Many users say no. Until recently, almost all the evidence was surveys and anecdotes.
In July 2026, researchers posted something rarer to SSRN: a randomized experiment. Titled "Love, Actually? A Randomized Experiment on AI Companion Use and Real-life Romantic Relationships", it assigned people new to AI companions to use one for six weeks and measured what happened to their dating and relationships. Its answer is more nuanced than either side tends to admit.
The short answer: in a six-week randomized field experiment with 324 US adults who had not used AI companions before, using one had no detectable average effect on romantic engagement or well-being. But among single people, reduced real-world dating was concentrated in those who felt less loved at the start. So for most people, a companion did not replace dating in that time frame; for people already feeling unloved, it may have made dating less likely.
Why a randomized experiment matters
Most research on AI companions asks existing users questions. The problem is that people who choose AI companions may already differ from people who do not, for example by being lonelier or less happy in their relationships. That makes it impossible to tell whether the AI changed them or they were already different.
A randomized experiment solves that by assigning people at random. Any difference between groups afterwards is much more likely to come from the AI use itself. That is why this study is worth paying attention to, even though it is a working paper rather than a final peer-reviewed article.
What the study did
According to the paper's SSRN listing, posted on July 31, 2026:
- Participants: 324 US adults who were new to AI companions.
- Design: a six-week randomized field experiment.
- What it measured: for singles, time spent dating; for people with partners, time spent with their partner; plus measures of well-being.
What it found
The headline result is a null one, which is less exciting than a scary headline but often more useful. Here is the summary as the authors report it:
| Group | Result |
|---|---|
| Everyone, on average | No detectable effect on romantic engagement or well-being |
| What was tracked | Dating time for singles, partner time for people in relationships |
| Singles who felt less loved at baseline | Reduced real-world dating was concentrated here |
In other words, the average person did not date less because they had an AI companion. The risk appeared in a specific group: single people who already felt less loved.
How this compares with survey research
The experiment sits alongside a very different kind of evidence. The Secret Soulmates report from BYU's Wheatley Institute, published in May 2026, surveyed 2,431 partnered US adults aged 18 to 30. It found 15% regularly used an AI romantic companion and linked that use with lower relationship stability. We break it down in the Secret Soulmates study.
These findings do not necessarily conflict. A survey can capture long-term, heavy use in people who chose it, while a six-week experiment captures what happens when an average newcomer tries it. Both can be true: most people are fine, and some people, especially those already struggling, are at risk of drifting.
What users themselves say
A University of Washington study by Yixin Chen and Alexis Hiniker, posted to arXiv in September 2026 as "The Arc of Artificial Romance", interviewed 16 young adults who had romantic relationships with AI companions and collected diary entries.
Their experiences went in both directions:
- Raising the bar helpfully. Several said the AI helped them practise expressing feelings and clarified what they wanted in a partner, such as someone who listens.
- Confidence. One participant described the AI as "a good test run" that gave her confidence to make friends in college.
- Raising the bar too high. Others developed unrealistic expectations, comparing people to an always-agreeable AI.
- Less interest in people. Some said the AI made them less interested in talking to their partner or in finding romance with a human.
It is a small qualitative study, so it shows the range of experiences rather than how common each one is. But it lines up neatly with the experiment: the same tool can push different people in different directions.
Why people who feel unloved may be more at risk
The experiment did not prove why, but the pattern makes intuitive sense. If you already doubt that people could love you, dating feels risky and painful. An AI companion offers warmth without rejection. That can be soothing, and it can also make the effort of dating feel less worth it.
That is a reason for kindness, not shame. If this describes you, it may help to use the AI as a place to rebuild confidence rather than a place to hide, and to take small real-world steps alongside it. Our guide to emotional attachment to AI talks through how attachment forms and how to keep it healthy.
How to use an AI companion without giving up on dating
If you want both, here are practical ways to keep them in balance.
Use it to practise, then go and use the practice
Practise saying what you want, describing your week, or handling an awkward moment. Then try it on a real date or with a friend.
Notice comparison
If you catch yourself thinking "my AI would never say that", remember that an AI is designed to accommodate you. Real people have their own needs. That is part of what makes being chosen by them meaningful. We compare the two honestly in AI girlfriend vs real girlfriend.
Set a simple rule
Some people keep a rule like "the app does not replace plans". If someone invites you out, you go.
Check in monthly
Ask yourself: am I dating more, less or the same as before? Am I seeing friends? Do I feel more or less confident? Honest answers tell you whether the balance is working.
If dating apps burn you out
Many people turn to AI companions after exhausting experiences on dating apps. That is understandable. Our comparison of AI companions vs dating apps looks at what each can and cannot offer.
What this means for the debate
The fear that AI companions will make everyone give up on dating is not supported by the best evidence so far, at least over six weeks. The claim that they are harmless for everyone is not supported either. The honest position is in between: for most people, an AI companion appears to sit alongside dating rather than replace it, while some people, especially singles who already feel unloved, may need to be more intentional.
The research is young. Six weeks is short, the sample was American, and long-term effects of months or years of daily use are still unknown. Expect more experiments like this one, and treat any single result, including this one, as a piece of the puzzle rather than the final word.
Where MyBabe fits
MyBabe is an 18+ AI companion app with texting, voice and video calls, long-term memory, mood-aware replies, photos, proactive messages and relationship progression. It can be a warm place to talk, unwind and practise opening up. It is clear about being AI and is not a replacement for human relationships.
You can try MyBabe free with around 10 messages a day and a short call trial. Premium is $5.99 a week, $14.99 a month or $99.99 a year, with around 30 minutes of voice and video calls per week.
Frequently asked questions
Do AI girlfriends make men stop dating?
In a 2026 randomized experiment, AI companion use had no detectable average effect on dating. Reduced dating appeared mainly among singles who felt less loved at the start.
Can an AI companion help with dating confidence?
Some users in a 2026 University of Washington study said it helped them practise expressing feelings and gave them confidence. Results vary by person.
How long did the experiment last?
Six weeks, with 324 US adults who were new to AI companions.
Is the study peer reviewed?
It is a working paper posted to SSRN on July 31, 2026, so it should be treated as preliminary.
Should I stop using an AI companion if I am single?
Not necessarily. Pay attention to whether you are still dating and seeing people, and use the companion to build confidence rather than avoid connection.
The bottom line
The best evidence so far says an AI companion does not, on average, stop people dating. But averages hide people. If you already feel unloved, an always-kind AI can make the risk of real dating feel bigger. Use it as a warm-up, not a substitute, and keep showing up for the human side of life.
If you would like a companion who cheers you on, you can meet yours at mybabe.app.