It turns out the answer is not so clear. Empirical studies show that while choice reflects preference, it also influences it. For example, when given two identically appealing items, our brains often choose an arbitrary tie-breaker, yet have post-hoc rationale for why one was better. This rationale (or ir-rationale) is used the next time similar options to those two are presented.
Why do this? Because for the brain, time is energy, and energy is limited! Our brain consumes about 20% of our whole body’s energy. If we spent a lot of this energy trying to figure out a stalemate, we would be in trouble! Therefore, the most energy-efficient thing to do is to choose arbitrarily, or better yet have someone else nudge you.
This is where recommendation systems come in. The role of a recommendation system is to help us navigate the endless sea of choices. In evolutionary settings, it is beneficial to observe and copy others’ choices, a big part of human success has been our ability to learn from others in the tribe. We are biased towards listening to advice before making decisions. So, how convenient is it that we each now have our very own personalized tribe, a wise, all-knowing algorithm whispering in our ears?
In most cases, it’s great. However, there is a hidden downside. Since our preferences and choices are intertwined, recommendation engines not only help us make decisions but ultimately shape our preferences as well. Research by Jinjing Zhang’s group finds that even when recommendations were given out randomly, they ended up shaping listeners’ preferences. People are hard-wired to believe that a personalized recommendation has truth in it, that it would be silly to disregard it. In fact, the effect is so pervasive, that even when subjects listened to the music firsthand, the randomly given ratings from the fake recommendation engine still influenced preference.
This gives recommendation engines immense power, they can arbitrarily change your preference! It’s time to update the old adage of “If all your friends jumped off a bridge…” to “If all your search results are telling you to buy this company’s product…”
So, what is the solution? Clearly, recommendation engines are useful, in fact, they are needed to clear out the noise in the seemingly infinite set of choices we are presented with. I think the answer is to be more mindful of influence and not let the recommendation engine lead you blindly. And then, every once in a while, break the mold and make decisions from your own experience. Eat at a restaurant before looking at the reviews, and watch a movie without looking at the ratings. Sure, some of these experiences may end up being less than ideal, but you may end up discovering hidden gems, or better yet, you may expand your taste to new horizons the algorithm wouldn’t dream of.
https://sloanreview.mit.edu/article/the-hidden-side-effects-of-recommendation-systems/