ChatGPT-3’s debut in late 2022 shook up tech–and online dating was no exception.
Within weeks, Tinder users were reportedly penning bios with the AI tool’s chatbot interface.
In February, Mashable reported that OkCupid has used ChatGPT to generate questions for its users.
Some have voiced skepticism about the prospects of AI-enabled romance.
Psychologist Andrew G. Thomas found that ChatGPT overstated the similarities between men and women when it comes to what they value in a partner.
And an editor reported in Business Insider that her ChatGPT-authored responses on Hinge landed with a dull thud: “No one wrote back.”
The March 14 release of ChatGPT-4 sparked renewed interest in the platform.
CNN, the BBC, and other news outlets released stories on the launch.
That’s very different from the situation last year when early excitement over ChatGPT-3 in Silicon Valley was not initially matched by coverage in the legacy media.
According to OpenAI, the latest version of ChatGPT significantly outperforms its predecessor in terms of reasoning and accuracy. It is also, in their words, “safer.”
“GPT-4 is 82 [percent] less likely to respond to requests for disallowed content and 40 [percent] more likely to produce factual responses than GPT-3.5 on our internal evaluations,” OpenAI’s webpage on ChatGPT-4 states.
It further claims the platform produces “safer and more useful responses.”
AI safety has become embroiled in controversy, partly because of concerns about bias in GPT.
In a Twitter post from early December 2022, just after ChatGPT-3 was released, venture capitalist Marc Andreessen asserted that AI safety is equivalent to “AI censorship.”
“They’re the same thing,” he wrote.
“I am begging you to read the literature,” responded Miles Brundage, the head of policy research at OpenAI.
As with the launch and gradual reaction to ChatGPT-3, the rollout of ChatGPT-4 marked a promotional opportunity for those in online dating.
In a March 14 Twitter post, Jake Kozloski, CEO and co-founder of the dating app Keeper Dating, showed how his company is working to automate matchmaking with the new tool.
In an accompanying video, the AI system assessed the compatibility of a fictional man and woman.
It analyzed information about both of them that the user provided.
GPT-4 compared their preferences, teasing out similarities that suggested the two could be compatible.
Yet, it also identified gaps in the data it needed to judge the prospective partnership.
It requested additional information, so as to evaluate the match more comprehensively.
In a March 17 interview with The Epoch Times, Kozloski explained that Keeper had already implemented GPT-3 before GPT-4 made its appearance.
GPT-4, he said, is “more capable for our use case.”
Keeper’s embrace of cutting-edge technology to build serious, long-term romantic partnerships is in keeping with a seemingly paradoxical vision: It’s at once future-oriented and moored to some of the oldest traditions in human connection.
Kozloski co-founded Keeper with pseudonymous Internet reactionary Indian Bronson, whose Twitter bio states that he’s “fixing the birthrates” at Keeper.
Kozloski elaborated on that goal in Keeper’s “master plan:”
“We’re in a fertility crisis fueled by a marriage crisis that jeopardizes the future of humanity itself.
“Our mission is to address this crisis head-on by reversing the decades-long trend against marriage and family formation.”
Collapsing birth rates in the United States have drawn the attention of SpaceX founder Elon Musk.
“We just need to celebrate having kids,” Musk wrote on Twitter in response to a graph showing a collapse in U.S. fertility since the early 1970s.
Births have also cratered in Europe, Northeast Asia, and much of the rest of the world.
To move forward, Keeper looks backward to the kind of supervised matchmaking once common in the West and still prevalent in India, Africa, and other traditional societies.
Kozloski didn’t come to his tradition-oriented perspective from a conservative background.
“I grew up with divorced parents in Brooklyn. Most of my upbringing was very progressive in terms of values, so I wasn’t exactly indoctrinated into this way of thinking (if anything, the opposite),” he said.
He and his girlfriend are in a stable, monogamous relationship–the unquestioned norm for past generations, but less and less so for Millennials and Gen Z.
“I do believe it is the ideal outcome. I recognize that it doesn’t work for everyone, but I think it does for the vast majority of people far more than [those] who are actively in one,” he said.
Like any matchmaking aunt from the old country, Keeper is picky. It only matches users with partners who meet every single one of their criteria.
There’s no limit to the number of criteria people can list–”but obviously the more they provide, the harder it is to match them,” Kozloski said.
Keeper has made 40 matches so far; that’s 80 people total if you’re doing the math at home.
Ten percent of those matches led to long-term relationships.
According to Kozloski, interest is much broader.
“Our signups are in the low thousands,” he said.
Yet, while users have the option not to pay until they wed a match by Keeper, those who sign up must also be prepared to pay thousands of dollars.
“We are a white glove service today,” he said.
Kozloski hopes GPT can help drive down the cost of matchmaking, allowing them to expand their customer base and offer lower prices.
Kozloski hopes his matchmaking can be fully automated as GPT continues to improve.
“Today, we have matchmakers working with the [algorithm and] AI,” he said.
In AI jargon, they want to get that human out of the loop.
Keeper aims to create the first algorithm that makes genuinely good matches, rather than the inexact pairings on offer from existing services.
Kozloski thinks the first company to manage that feat would be to online romance what Google is to search.
“We plan on that being us,” he said.
If Keeper pulls it off, they may have ChatGPT to thank.