Choices and Algorithms
The previous blog talked about how studies and predictions have probabilities and assumptions associated. As pointed out by my friend Hrisheekesh one of the main reasons for ignoring probabilities and assumptions is that people are looking for actionable insights. Everyone has to make hard choices, even though there could be uncertainties about the future. This blog is about these choices and how algorithms are impacting them.
Making hard choices requires evaluating the probabilities, assumptions in detail. But with the rapid advancement in technology over the past couple of decades, there has been an explosion of information and choices. From buying stuff to watching videos, we are inundated with options. These options can confuse us, or rather they are meant to confuse us. So we are left with two options to simplify our lives — 1. reduce the number of choices or 2. reduce the time spent on each choice in evaluating them. By making conclusions from the media headlines, we often resort to the second option. But are there ways to opt for option 1 to make our lives easier?
We have been reducing the number of choices over the years — by being loyal to a brand during shopping or reading certain newspapers or authors. But this reduction often depended on our own preferences and volition. But things have changed with technology. Companies are often reducing our choices and under the guise of personalization using algorithms. As I have heard from one of the experts in Analytics, algorithms are making the choice, but giving an illusion to the user that s/he is making the choice! Amazon is recommending us what to buy. Facebook is telling us which news may interest us and putting them higher up in the feed. While this gives us the illusion of making better choices, it faces the same problem I mentioned in the previous post. These recommendations are based on assumptions and probabilities! But this time from the company which is shortlisting options for you.
The problem is not the recommendations per se. It is always good to have a helping hand. Don’t we often ask our friends/ family for recommendations? But I see two problems in being over-reliant on algorithms. First, better alternatives are deliberately kept out of our reach just because algorithm “predicts” so, which forces us to choose an inferior alternative. For example, Amazon showing only certain products, because they are tailored to you (and hence, likely to be sold). Second is the amplification of the biases! Feeding with only things you have done historically reinforces biases (this is making the society increasingly polarized, but it is topic for another blog). If I read certain books or buy some products does not necessarily mean that I will continue to do so. We are missing out on the joy of discovering new things!
So how can we make the process of choosing easier? First is, of course, being aware that what algorithms are recommending may not be best for you and venturing beyond the suggestions. Other option, which is better in my opinion, is to take back the choice in own hands through curated content. Though the accuracy of algorithms has improved considerably, humans can provide an additional layer of sophistication in recommendations. So rather than Twitter telling you which are the best articles to read, why not follow someone who has put that effort into reading articles and posting their links. Use curated travel sites like Lonely Planet or Rough Guides rather than just Googling for travel plans! Don’t we often watch stuff on Netflix based on our friend’s suggestions, even though Netflix continues to show us recommended content? Or don’t we ask friends n family while buying a mobile to reduce our choices?
In my opinion, while the algorithms continue to improve and predict better, they use machine learning algorithms and strive to improve prediction without understanding underlying factors. I believe that the curated content and recommendations by humans will grow in future. While algorithms will continue to provide the first level of shortlisting, the last mile of selection will be done by humans. Who knows, there could possibly be professionals in the future, who will guide us in shortlisting our choices to make better decisions!