Why Don’t Dating Apps Work for so Many People?

by Brittany Vincent a day ago 5 min read

Why typical dating apps are broken (and how AI can actually fix them)

If you’ve ever felt like dating apps are a second job that rarely pays out, you’re not imagining it. A huge amount of time goes into swiping, chatting, and “optimizing” your profile, and very little comes out in the form of real, mutual, exciting connections.

From Iris’s perspective, the core problem isn’t you. It’s the way most apps are designed and how their algorithms work. Once you understand that, you can also understand why AI, used differently, can finally change the experience.

Why don’t dating apps work for so many people?

Most dating apps were built as engagement products, not as “mutual attraction engines.”

Their incentives are simple:

Keep you in the app as long as possible
Maximize swipes, taps, and time-on-screen
Show just enough “wins” (likes, matches, messages) so you don’t quit

That leads to a familiar pattern:

A small group of highly visible profiles gets most of the attention
Many people feel invisible, even if they’re genuinely attractive in real life
You get flooded with low-intent matches, dead chats, and first dates that go nowhere

From Iris’s research, the deeper issue sits in the math of attraction. Strong attraction to a random profile is rare, on the order of 1 in ~1,000 on one side. Mutual strong attraction between two random people is closer to 1 in ~1,000,000. When apps treat everyone as equally likely and just throw more profiles at you, they’re fighting against that math, not working with it.

What is popularity bias in dating apps?

Most swipe-based apps use ranking algorithms that boost profiles getting the most attention: more likes, more right-swipes, more messages. That creates popularity bias:

Profiles that start out getting more attention are shown even more
Everyone else is quietly pushed down in the deck
The “rich get richer” while others get fewer impressions and matches

For users, the experience looks like this:

Some people get overwhelmed with shallow interest they don’t really want
Others barely get seen, no matter how compatible they might be
The pool you see isn’t “everyone near you,” it’s “who the algorithm decided deserves attention”

This bias also distorts your sense of reality. You might feel like “everyone” wants the same handful of people, or that your own type is “unrealistic,” when in fact you’re just seeing a narrow, algorithmically boosted slice of the population.

Iris was built explicitly to move away from popularity bias. Instead of asking “Who is popular?” the focus is “Who are you strongly attracted to, and who is likely to feel that way about you?”

How does catfishing distort results?

When verification is weak or optional, a lot slips through:

Old photos that no longer match reality
Heavy filters and edits that change someone’s real appearance
Fake or stolen photos
Bots or low-quality profiles that are not genuinely there to meet

This doesn’t just waste time; it contaminates the data about attraction:

You might swipe “yes” on a face that doesn’t exist in real life
You might swipe “no” because you’re skeptical, not because you aren’t attracted
Over time, the app learns from distorted signals and gives you worse recommendations

From Iris’s point of view, any serious AI for dating has to start with authentic profiles. If the inputs are fake, the outputs will be useless. That’s why Iris takes a harder stance on verification and authenticity than a typical swipe app: genuine people, real photos, and clear intent are treated as non-negotiable.

Why does swiping create addiction without results?

Swipe interfaces are designed around variable rewards, the same pattern used in slot machines:

Most swipes feel like nothing
Occasionally you get a small “jackpot”: a match, a like, a message
Your brain gets a hit of dopamine and you keep going, hoping for another win

The result:

You can swipe for an hour without meaningfully improving your chances of a great date
Matches become quick hits of validation, not curated opportunities
The app feels busy and exciting, but your real life doesn’t change much

Iris’s internal narrative treats this as a structural problem: if the metric is “number of swipes,” the product will be optimized for more swiping, not better introductions. That’s exactly what AI needs to reverse.

How can AI focus on quality over quantity?

AI can either amplify the old model (more engagement, more swiping) or build a new one. Iris chose the second path.

Done right, an AI dating app behaves less like an arcade and more like a match analyst that learns your personal “AttractionDNA”:

You train the model on real examples
- You show the app people you find genuinely attractive
- You also show clear “no” examples
- Over time, the AI learns patterns in faces, expressions, styles, and more
The AI searches for strong, not lukewarm, attraction
- Instead of “anyone in your area,” it prioritizes profiles that fit your pattern of strong attraction
- It also looks for signs that the other person is likely to feel that spark back
The system works with the math, not against it
- If strong one-sided attraction is ~1 in 1,000, AI can narrow the pool dramatically before you ever swipe
- Mutual strong attraction (the ~1 in 1,000,000 scenario) becomes something the system is actively targeting, not leaving purely to chance

Iris is one concrete example of this philosophy: it uses AI not to write your messages or gamify your time, but to predict mutual attraction and surface fewer, better matches.

What outcomes should I expect with AI-driven matching?

No responsible product can guarantee that you’ll find “the one.” But an AI built around mutual attraction and authenticity can reasonably aim for improvements like:

Fewer, more relevant profiles You see a smaller set of people that are closer to your actual type, instead of endless noise.
Less time swiping, more time deciding The work shifts from “hunt through thousands of profiles” to “decide whether these few, highly curated options are right for you.”
More matches that feel exciting on both sides Because the system is looking for mutual attraction, you’re less likely to end up in lukewarm, one-sided situations.
Lower emotional burnout When you’re not constantly chasing small hits of validation, it’s easier to stay grounded, hopeful, and selective.

In Iris’s case, the entire product is oriented around those outcomes: use AI to raise the quality of intros rather than inflate the quantity of distractions.

How do verified intros change the experience?

“Verified intros” combine two ideas: real people and mutual intent.

In practice, that means:

Profiles go through meaningful verification so you know there’s a real person behind the photos
AI has already done a layer of filtering for potential mutual attraction
You’re introduced only when there’s a strong reason to believe both of you might genuinely click

This changes the feel of the app:

Less anxiety about whether the person is real or what they actually look like
Fewer situations where you invest in someone who was never serious to begin with
First messages that feel more intentional, because both sides already passed multiple “filters”

Iris was built from the ground up around this concept. Its verification stance and AI matching work together: authenticity + attraction, safety + quality, not one or the other.

Bringing it all together

If dating apps have felt broken for you, it’s not a personal failing. It’s the result of:

Popularity-driven algorithms
Minimal verification and catfishing
Swipe mechanics that reward time spent, not real connections
A total disregard for the actual math of mutual attraction

AI doesn’t magically fix that on its own. But when it’s used to learn your preferences, predict mutual attraction, and gate introductions to verified, authentic people, the experience changes.

Iris is one example of an app built on that new model: fewer swipes, more signal, and a system designed to make rare, mutual attraction a little less of a lottery.