My recent foray into the online dating world using the Bumble and Hinge apps got me thinking about artificial intelligence and algorithms. If my experience is indicative of the power of AI, I suspect our fears of being tricked, manipulated and ultimately ruled by these systems is over-hyped.
Both of these apps offer basic filters for me to indicate my preferences in potential dates. Filters include age, location,, sex, height, smoking, alcohol use, drug use, religion, children, exercise frequency, race, etc. Interestingly, neither app allows you to select body type as a filter. I guess that’s not PC anymore – even though I know of nobody who doesn’t consider body type and looks when dating.
Then each app purportedly “learns” my preferences as I accept or reject presented candidates so it can better present me with better matches.
I entered in my basic filters parameters and began swiping. At first, I was highly selective, accepting only candidates who met my 1% criteria. Out of several hundred candidates, there were less than 10.
I rejected many people. I rejected women who:
- Were overweight (frequently morbidly obese)
- Smokers and heavy drinkers
- Had small children
- Mentioned their #1 priority was being a mom
- Said they wanted kids
- Wanted lots of fine dining and international travel
- Expressed a desire for adventures and weekends getaways
- Wanted a man who was outgoing, extraverted and interested in partying
- Were ugly
- Were enthusiastically opinionated and bitter about politics or vaccines
- Wanted to do a lot of hunting, camping and fishing
- Displayed pictures that looked so different from one another that I couldn’t believe they were all the same person
- Expressed an unusually strong interest in food (more common than I expected)
- Were too old, too young, too far away
- Were lesbians. transsexuals, bisexuals, pansexuals, men and transvestites
- Wanted a man with long hair and tattoos (clearly not me)
These are all fairly basic criteria. Some of them map directly to the filter preferences that are able to be set in the apps.
And yet, now a week into this, I keep getting presented with many candidates who fall into these undesirable (to me) categories.
In fact, I get presented with people I specifically rejected. Over and over.
If this is an example of the power of omniscient AI and algorithms, we have no worries.
Then I thought about the automated customer service chat bots I’ve attempted to use online when I need customer support from various retailers, manufacturers and agencies.
They can never help me with what I need.
I’d guess that 90% of the time, my service request is not unique. I need to return a product, get support, check on a status etc. The bot then sends me to the sales page. Or gives me instructions to check the web site. Or sends me to the FAQ which does not answer my questions.
Another AI failure.
I always end up having to call someone on the phone to get a simple resolution.
When I think of AI and algorithms, I imagine a system that “learns”. But all we really need is a good database of common answers that filters and sorts. I could probably write one myself using my rusty BASIC coding skills.
I have a feeling that the dating apps algorithms are working exactly the way they were designed to work. They present an occasional “hottie” interspersed among many undesirable candidates. That way, they keep users clicking and sorting on the off chance they get lucky and hit on a winner.
I suspect what they purport to do – find the best matches for each individual – is simply typical marketing bullish**.
All I really need to do is frame this using one technique – Follow the Money. The money continues to flow to dating apps as long as users keep their monthly subscriptions active. If we actually found good matches and built relationships, the apps would lose their customers.
It’s good to keep this in perspective.
As for chat bots, automated customer service and similar AI support tools, the same axiom applied. Follow the money. It is cheaper to frustrate customers into giving up than it is to provide good service. It’s even cheaper still to do the with an automated software tool than pay a human being to do it.
So I don’t fear a Terminator-like end of world scenario from AI and algorithms. More realistically, I can envision a future somewhere between Idiocracy and WALL-E.