The San Francisco edition: the city that built the current AI boom still can't get its own dating market to add up
Every city in this series has had some claim to AI relevance — one hosts it, one funds it, one trains it, one manufactures the chips underneath it. San Francisco doesn't need to make a case. This is where the current AI boom actually started: OpenAI, Google DeepMind's largest research org, Meta's AI teams, and a dense cluster of the frontier labs building the models the rest of this series keeps referencing are all headquartered within a few square miles of each other here, alongside the venture capital that funds nearly all of it. If any city on Earth has both the technical capability and the cultural instinct to solve dating with an algorithm, it's this one. It hasn't. And that failure is more interesting than any success would have been.
Because San Francisco's dating problem is, by now, one of the most publicly litigated in the country, and the popular explanation — a lopsided gender ratio driven by tech's hiring patterns — turns out to be only part of the real story. Yes, San Francisco does skew more male in its 20s and 30s dating-age population, and the tech industry's hiring patterns are widely blamed for it; one recent event, an "Enforced Ratio" party thrown specifically to counteract the imbalance, drew 200 women and 200 men and made local news simply for managing 50/50. But a deeper analysis of Census data complicates even that: filter for college-educated singles specifically, and the ratio flips in the opposite direction in people's 20s, before the gap reopens again by the 40s. Neighborhood adds another layer entirely — the Mission, SoMa, and the Castro run more male, while the Sunset, Russian Hill, and Presidio Heights run more female. That's at least three different, individually accurate versions of "the ratio," depending on age, education, and zip code — precisely the kind of layered nuance a single national algorithm has no real way to hold at once.
But per local matchmakers who work this market daily, the ratio was never actually the core issue. The deeper pattern is cultural: a city of famously accomplished, analytical people who apply the same optimization instinct to dating that they apply to everything else, and get stuck in exactly the way you'd expect. As Luvo Matchmaking's own read on San Francisco puts it, this is a city where "the best" is often associated with precision — but matchmaking isn't something you can fully optimize, and two people can check every box on paper and still stall out because they never leave evaluation mode. That's not a data problem. It's the direct product of treating dating like a system to be solved, in the one city on Earth most inclined to do exactly that.
Now compare that instinct to what SF-built AI is actually excellent at, which is the same logistics category this series keeps returning to. Nationally, Phocuswright found 56% of U.S. leisure travelers used AI to plan at least one trip this year, and 78% of people who tried it booked something based primarily on its recommendation — a clean, bounded, well-instrumented problem, exactly the kind SF's frontier labs are best at solving. SFO sits inside the metro that built the actual models doing that optimization for the rest of the country.
Matchmaking has never behaved like that kind of problem, and the industry's own reckoning makes the point without needing an SF-specific example. Justin McLeod, who built Hinge into the country's most dominant dating app, left to launch an AI-first matchmaking startup backed in part by Match Group — and has said plainly that even his new venture treats AI as a coach, not a decision-maker. Three Day Rule, fifteen years into human-run matchmaking before layering in an AI-matching product, drew early criticism for a specific failure: both sides of a match receiving identical AI-generated opening lines, turning the actual humans into spectators in their own conversation. That's a particularly unconvincing failure mode in the one city whose entire population can recognize a generic, over-optimized pitch on sight.
None of this argues AI has no place in San Francisco dating — coaching tools that sharpen a profile or catch a self-defeating pattern remain a legitimate, evidence-backed use case, distinct from trying to automate the actual judgment call. But San Francisco might be the single clearest proof point in this entire series: the city with the deepest concentration of frontier AI talent and capital on the planet, and still one where, per Luvo's own read, "great matches aren't just about compatibility — they're about alignment beyond analysis," something no model built here has cracked for its own hometown. That's the case for Luvo's San Francisco practice, sourcing introductions from real communities across SoMa, the Mission, and Pacific Heights rather than a database that can score compatibility but can't get two people to stop evaluating each other long enough to actually connect. San Francisco will keep building the AI that optimizes everything else in the world. Getting its own residents out of evaluation mode is still, even here, a human job.