The art of matchmaking has traditionally been the province of grandmas and best friends, parents, and even—sometimes—complete strangers. Recently they’ve been replaced by swipes and algorithms in an effort to automate the search for love. But Kevin Teman wants to take things one step further. 

The Denver-based founder of a startup called AIMM has built an app that matches prospective partners using just what they say to a British-accented AI. Users talk to the female-sounding software to complete a profile: pick out your dream home, declare whether you consider yourself a “cat person,” and describe how you would surprise a potential partner.

At first glance, that doesn’t seem too different from the usual swiping-texting-dating formula of modern online romance. But AIMM, whose name is an acronym for “artificially intelligent matchmaker,” comes with a twist: the AI coaches users through a first phone call, gives advice for the first date, and even provides feedback afterwards. Call it Cyrano de Bergerac for the smartphone era.

“It’s guided,” Teman says. “You’ll say if you’ve gone on your date, and it gets feedback on chemistry and how you felt overall. If you indicated highly, you follow up and set up a new date. If it didn’t go well and you like the person, it says to give it some time and be patient.”

If the attraction is one-sided, the app might gently prod you to “move on,” though Teman insists it is gentle, sensitive, and subtle. “I mean, it won’t say to the guy ‘The girl doesn’t like you’ directly,” he says.

Teman hopes AIMM can disrupt the $2.5 billion online dating industry, currently dominated by the Match Group owned by IAC, whose conglomerate includes Match, Tinder, OkCupid, and Plenty of Fish. Once considered an odd way to find a partner, online dating has grown to become the most common way for couples to meet, according to a recent study by researchers at Stanford and the University of New Mexico. Nearly 40% of heterosexual couples met online in 2017; for same-sex couples, the figure was 65%.

For all that popularity, dating apps are plagued with complaints, from ghosting to fake accounts to reports of harassment.

“Finding love is the most important thing for people,” Teman says, citing his own difficulties navigating online dating. “The dating industry is broken.”

Over the last few years, use of voice technology has skyrocketed—one in five Americans now owns a voice assistant. Over 2.5 billion of them are in use today, and experts think that could more than triple to 8 billion by 2025.

AIMM is not the first to put voice technology at the foundation of an online dating service. Last summer, Match announced it had partnered with Google to create a dating advice chatbot named Lara, which pulls a profile a day and, if there’s mutual interest, offers advice on where to grab a drink, what someone’s all-important first text should be, and how to structure the second date.

AIMM, which Teman says came out a year before Lara, is similar in its soothing encouragement (as soothing as an AI can be, at least). Teman argues that his app is much simpler to use—Lara requires a Google Home device and the Match app, while AIMM is all done on a smartphone. 

“It coaches you on what to say on the [first] call,” he says (AIMM discourages communication beyond the single call before the first date). “Some of it will encourage you to be calm. Some will give you specifics into what kind of person they are, like ‘traditional’ or ‘modern’ lifestyles.”

Those clues are meant to help the user figure out how to talk to a date. For someone with a more “traditional” lifestyle, a walk in the park might make more sense, for example. A “modern” lifestyle could make a rock-climbing date a better fit.

But as modern and high-tech as AIMM itself is trying to appear, it comes off as way behind the curve of user behavior—even sexist. The app is designed for a man to ask a woman out, for example, and the AI feeds only the man information and suggestions for a first date based on mutual interests. That seems like an oversight, at best, for an app that will compete with the likes of Bumble (in which women must make the first move).

It’s also set up to favor heterosexual couples—which is at odds with the findings from the Stanford study, which show members of the LGBTQ community are much more likely to use online dating apps. Teman acknowledges that AIMM’s inherent design isn’t as LGBTQ-friendly, but says he’s making tweaks based on “homosexual users who reached out to me ... suggesting other things that I could add that are more about the homosexual lifestyle.”

When asked about what these suggestions specifically were, Teman was vague: “They said that the questions seemed like they were all for straight people and there were no questions about Pride lifestyle. So I added some things about—if you’re homosexual, it can go into some questions that are specifically about your lifestyle.”

Andrew McStay, a professor of digital media at Bangor University in the UK and author of Emotional AI: The Rise of Empathic Media, says he believes AI is the future of dating and considers voice-based apps potentially more “natural,” but he thinks we’re still in the technology’s infancy.

“This is less a dating question than a human-device interface issue,” he says. “There’s every reason to believe that people will increasingly interact with devices and content through voice, especially as natural-language processing techniques improve.”

There may, however, be an ethical issue in AIMM’s design, McStay points out. When a first date is arranged, not only does it reassuringly coo, “Don’t worry. I will be with you every step of the way, and I will try to make this an effortless process for you”—it also gauges your reaction. The PR video shows AIMM observing, “I can see by your smile that you feel good.”

“There’s a video component analyzing your emotions,” Teman said when I asked him how the app could know if a user was happy or not. To McStay, that’s where the interesting future of AI in dating lies.

“This portends a lot in terms of the next stages for such services,” he says, citing prosody, or the various inflections and intonations humans take on when speaking, to “gauge voice-based enthusiasm and interest levels of users.” A user might gleefully exclaim “I’m interested!” when matched with a person, for example, or mumble “I’m interested” after a tiring day at work. McStay said that the latter might be misinterpreted as lack of interest.

But he says that analyzing facial reactions to gauge a person’s interest is potentially problematic: “One has to pay careful attention to the method of inferring emotion from faces or voice.”

Take, for example, “resting bitch face”—the appearance of ennui, slight disgust, and “bitchiness” that might not at all reflect what the person is feeling and has been tied to sexism in the workplace (a man with such an expression might be read as “serious,” but a woman with the same expression is often read as “aloof”). 

Computers are even worse at interpreting emotion on the basis of people’s expressions. Emotional nuance requires human input, context, and history—something machine learning simply can’t do, at least not yet. 

Whether or not AIMM even works, let alone improves on what other dating apps have to offer, is still very much up for debate. Thus far, Teman says, it has set up several dates—but none have stuck yet. In the coming months, AIMM will offer users the option to be paired with a human matchmaking service, with in-person sessions helping them through the process.

“Anything that helps them feel cared for,” Teman says. Right now, he adds, “daters are left cold in the dating app world.”