I’m wondering whether maybe “dating service” might be a genuine “killer app” for AI. I, myself, am an AI cynic, seeing that the hype and concomitant human folly have far outstripped the proven, solid uses for this new technology. However, perhaps human matching is actually a task an AI algorithm could successfully tackle.
There already are a few AI dating services out there, even after removing the chatbot girlfriend/boyfriend providers and the AI dating advice sites, but even the current AI matchmaking sites apparently still rely on questionnaires and so they don’t go far enough for what I am talking about.
My not-very-controversial thesis is that good dating is an interpersonal information problem, not just acquiring the information on potential candidates but also what to do with it. Using voluntary questionnaires has proved suboptimal, and frankly, letting the participants make choices based on the information provided has no special track record, either.
What if matchmaking is best accomplished by moving candidate consideration all the way into true pattern matching using abundant loads of data? One success story for AI that everyone likes to point to is medical image analysis and lesion spotting. What is that but machine-learned complex pattern matching? Maybe the information fields we humans both throw off and also need to have about potential partners can be analogized to a good CAT scan.
I am not talking about questionnaires here, or perhaps any voluntarily produced information, though there’s no reason to exclude that stuff. Perhaps our true personal contours are best revealed by the digital footprint we lay down every day, both voluntary and involuntary, both personal and demographic, both past and current. We each have limited purview over our data store and can’t really influence it or “fake” it. Each person’s full data store is quite large, but certainly AI can hoover it all up.
Then what? Once you have those millions or billions of huge personal-profile data troves, what do you do with them? What comparisons do you make and what algorithms do you follow? Do opposites attract? Does like-mindedness really promote compatibility? Who knows? We have never to date anecdotally produced good answers to those dating and compatibility questions. So, keep hoovering!
We have the Internet, and independently vast demographic records, not to mention evolutionary knowledge, at our AI disposal. So, let’s find out what all those data themselves tell us for how to go about finding those tumors, I mean, those successful matches. Let’s look at the history of successful togetherness (and perhaps more importantly, failed togetherness) and see what the ocean of data tell us. Anyone who has run a statistical “t test” and watched solid causative factors come out of seeming random splotches knows the magical feeling of organization rising from apparent disarray.
Sure, the Internet and all other records are wildly poor indicators of human romantic success, at least to our human eyes. We are talking tons of chaff per each small grain of actual reliable index to happy couple-hood. On the other hand, there is so much data that even if the ratio is a ton to an ounce, with enough grinding it may still produce a usable amount.
And of course, the patterns found from such peta-analyses may be not only beyond human intuition but beyond human comprehension. The proposed matches might be mind-boggling and foolishly implausible. But, it similarly does not matter how the medical-image AI analyzer finds the tumor, only that it reliably does. Even if the first few proposed matches were unappetizing or felt laughably foolish, still, the only way to know for sure is to try a few. And if some of those matches actually worked, that would produce high quality, focused data for moving forward.
Would it work? Who knows? Is it any worse than current AI slop from clearly inappropriate AI uses and crazily stretching to fit AI to everything? Hardly. All I can say for sure is that with this post I have just killed the seminal conceptual patent for AI dating by making this public disclosure. You’re welcome.
[link] [comments]