Dear OpenAI Team,
My name is Ruben Nakaidze, and I would like to share a product idea that I believe could have a meaningful social impact if developed responsibly with AI.
The idea is an AI-powered relationship compatibility system that could help people find partners based on deep compatibility rather than primarily on photographs, age, location, interests, or short questionnaires.
I believe there is a major problem with the way people meet today.
Millions of people want a serious relationship, love, marriage, or a family, but many repeatedly meet people who are fundamentally incompatible with them. They may have completely different values, emotional needs, communication styles, expectations, or attitudes toward loyalty and family.
Traditional dating applications mainly show people how someone presents themselves. But a profile does not necessarily show who that person really is.
My idea is to use AI’s ability to understand people through long-term conversations and interactions to build a much deeper, dynamic understanding of each user — always with explicit user consent.
For example, over time AI could learn that a particular person values loyalty above almost everything else, needs emotional warmth and affection, dislikes controlling behavior, wants a serious relationship and family, prefers calm communication during conflicts, and needs a partner who is emotionally mature and trustworthy.
This information would be much more meaningful than simply asking the person to select “loyal,” “family-oriented,” or “romantic” in a questionnaire.
One of the most important parts of my idea is that questionnaires alone are not enough.
People do not always answer honestly. Sometimes they do not even fully understand themselves. Their behavior, reactions, experiences, and repeated patterns over time may reveal things that they would never put into a questionnaire.
Therefore, the AI should distinguish between:
-
what a person says about themselves;
-
patterns consistently observed over time;
-
and how confident the AI is in each conclusion.
The system should never pretend to know a person with absolute certainty.
Once a user voluntarily chooses to participate in matchmaking, the AI could compare their compatibility with other consenting users.
The matching process should not simply ask:
“Do these two people like the same movies?”
It should ask something much deeper:
“Are these two people likely to understand, respect, support, and emotionally complement each other?”
Compatibility could be evaluated across areas such as personality, values, emotional needs, communication style, relationship expectations, lifestyle, family goals, boundaries, and previous experiences.
For example, if Person A independently tells the system that they are looking for a loyal, caring, serious, family-oriented partner, and Person B independently expresses similar values and has personality characteristics that complement Person A, the AI could determine that they may have a high level of compatibility.
However, privacy and consent would be absolutely fundamental.
The AI should never secretly introduce people to each other.
Both people would have to independently opt into this type of matchmaking.
If the system identifies a potentially strong match, it could simply ask both people:
“I have found someone whose relationship values, personality, and expectations appear highly compatible with yours. Would you like to consider meeting them?”
Only if both people agree would an introduction take place.
The private conversations of either person should never be revealed to the other.
The system should not say:
“She told me this about you.”
or:
“He said this about his previous relationship.”
Private information should remain private and only be used, with permission, to improve compatibility assessment.
The people should be introduced as people — not as two strangers who have been given access to each other’s private psychological information.
I also believe this technology should not be designed to maximize the number of matches.
The objective should be the opposite:
Fewer random matches, but potentially better matches.
The long-term vision is not simply another dating application.
It could become an AI-powered relationship compatibility layer that helps people better understand themselves and, with their consent, discover people with whom they may have a genuinely strong chance of building a healthy relationship.
Of course, no AI can guarantee love, marriage, or happiness. Human relationships are too complex for that.
But AI could potentially help solve one of the biggest problems in modern dating:
“There are millions of good people looking for meaningful relationships, but they do not know how to find the right person for them.”
I believe AI could help bridge that gap.
I am sharing this idea because I believe systems such as ChatGPT could eventually make something like this possible, provided it is designed with strong privacy protections, transparency, user control, and mutual consent.
I would be very interested to know whether OpenAI has considered anything similar, or whether this concept could potentially be explored as a future capability or research direction.
Thank you for taking the time to read my idea.
Sincerely,
Ruben Nakaidze