We often hear about algorithms that can forecast our preferences, suggest products, or even detect health issues before they become apparent. How can a machine, which doesn’t have consciousness or feelings, accurately predict something so complex as human behavior? I’m fascinated by this intersection of technology and psychology and would love to hear from anyone who can shed light on the mysterious ways these algorithms seem to ‘know’ us. What’s happening behind the scenes that makes this possible?
In: 10
Imagine if you had to calculate all the physics and calculus required to figure out the correct angle, speed and position of your leg every time you needed to take a single step. You’d get no where.
Your brain however ignores all that fancy number stuff and just does it because it knows.
What you’re asking about is basically the reverse of this. Humans do a lot of stuff unconsciously just like how I described. We don’t NEED to know the information, ourselves just act on it at times. These algorithms give physical rhyme to our reason. Basically they can explain what we’re going to do because they figured out the stuff our body intrinsically knows.
Kind of how you don’t know how or why your body pumps blood. You can learn it but you as a human don’t just come out knowing that information. Your body just does it. The same way we learned that we could track stuff in your blood as it flows around your body, someone learned that we could track your selections at a store over a few trips and now they know what you are most likely to do.
This gets extrapolated out the more data you have. If the store buys data from another group to find out more about your habits and the habits of others you can start making educated guesses. Oh this woman is focusing more on nutrition and stopped buying alcohol and cigs? I betcha shes pregnant. ect
They don’t know, but they have a pretty good idea.
correlation correlation correlation.
If someone buys X, they are more likely to buy Y, expand this to a whole line of products related to another line of products and you can find that people who buy protein powder are a lot more likely to buy gym clothes etc. etc. There’s no reason or thought, it’s just all in enormously massive amounts of data about human behaviour all calculated through to optimize advertising.
Algorithms don’t understand the intricacies. They collect a lot of data from a lot of people, and when the data contain patterns algorithms can find them. So for example a woman might stop spending money on alcohol and nine months later start buying diapers (obvious pattern; pregnancies last 9 months and alcohol is counter indicated for pregnancy) – that’s a very common pattern, shows up in a lot of women, and you can use it to predict that someone might start buying diapers.
A very common pattern is going to be a very good predictor because it’s going to fit a lot of people because it’s common. If you want to use it to understand human behavior, you’ll need to actually interpret why the pattern exists.
Also note that the pattern is never going to be a perfect predictor: there are going to be people for whom it’s predicting incorrectly. Those are the cases where there are intricacies that break the pattern.
Many others have commented on aspects of pattern recognition and large data samples.
I’ll also note that as humans, sometimes our actual behaviors are a lot different than what we think they should be. Observing “actual” behavior provides much better data than asking people how they think they ought to behave.
There’s plenty of examples, but my neighborhood is great at demonstrating this. I live in an affluent neighborhood, filled with plenty of “love is love, all lives matter, we’re all human, etc etc.” signs. How do you think the neighborhood voted when multi family, county supported housing was proposed? Likewise, at one point everyone vehemently supported speed humps to slow traffic and “save the children!” That is, until the plans were released and the jumps were in front of their house.
The short answer is that humans are not nearly as nuanced and intricate as we’d like to believe ourselves to be.
Latest Answers