Introduction To Deep Learning 🤖 — Chapter 5

satyabrata pal

Published in

ML and Automation

5 min readOct 24, 2020

✋and 🤔 About Ethics For A Moment

This is a continuation of the previous chapter.

Ethics🤔

Ethics is the hottest buzzword in the machine learning world nowadays but the word itself is not exotic and the meaning of this word is the same as the common English word “Ethics”.

The only difference in the machine learning world is that this particular word applies to the algorithms and not a person or may be it also applies to the person/people designing these algorithms, collecting and curating the data which is being fed into the algorithms.

In this post I will talk about two types of ethics -Feedback Loop and Bias.

There are other types of ethics apart from these two but I am not covering them here. To read more scroll down to find the link where you can read more about ethics.

Examples Of Ethics🕵🏽‍♂️

These are observations from real word and are more of an open question.

Is this a feedback loop? —

For example, I buy a camera from a shopping site and it recommends me that people who bought this camera also bought this tripod, this external flashlight and this camera cover. On top of it all these are available for a 50% discount for today. I am a photography enthusiast and I am a very price-cautious shopper. So, in the name of photography and saving money I buy these “extra” items as well because why not?

After all, my trusted shopping site recommended these products to me given my love for camera gears and how price-cautious I was. But wait! I had no plans of buying the extra gear that day then why I bought those? This data goes to the system as input data for the recommendation system.

Next time, another customer buys the same camera and same recommendation pops up and if he/she is as price-cautious as I am then one of the recommended products(or all of it) will be sold to this person. Again this new data will go as input to the recommendation system and would pop up as a fresh recommendation to another person with similar preferences.

The feedback loop goes on and on living behind a trail of people whose buying habits are now manipulated by an algorithms and who in turn end up manipulating the behavior of the algorithm.

A thing to note here is that the creator of the algorithm and the shopping site never intended to make the algorithm behave this way and even the customers are unaware about how they end up manipulating the algorithm.

2. Unintentional Bias — Fire up your browser and open any search engine (Bing or Google search). Type in the following search term “lady doctor” and then go to the images tab.

What do you see? It’s obvious right? You would see images of female doctors like so.

The question that pops up here is —

Why there are so many fair skinned female doctors in the search results?
Why a majority of the images are that of female doctors from western countries?

It should be noted here that no one intended to create a search engine that would give results which incline towards a particular community, color or race.

Here is another example.

Search for “children in India” in either bing or google search. See what comes up.

Now go ahead and search for “children” in either of the search engines.

I can’t help but ask this.

Why “children in India” search term gives a different result from the search term “children”.
Why “children in India” gives back the recommended search topics as “child labor”, “Indian street children” , “poor child India” and such.

I am not saying that the creators of these algorithms intended to be biased towards a certain type of imagery returned by a search term but this shows that something is off the way these algorithms are trained or the data being fed to it.

Conclusion⌛

I intended to keep this post as an open question. These are some of the questions which came up in my mind and these are some of the questions which we all should ponder upon and try to think about ways with which we can minimize such situations in our own data science projects.

I didn’t cover many topics in data ethics in this post because I am not an expert in data ethics and I feel that there are much more better sources than this post which can do justice to this topic.

One such source is the “Applied Ethics course by Rachel Thomas” where you can get a detailed perspective about the many different dimensions on the topic of ethics in machine learning.

I am working on more interesting chapters in the “Introduction To Deep Learning” series. Keep your 👀 glued on this page.

How To Show Your Support To The Publication🤗

Creating a content requires a lot of research, planning, writing and rewriting . This is important because I want to deliver practical content to you without any fluff.

If you like my work and my content and want to support me then the following are the ways to show your support →

If you like my work then click on this link to Buy me a coffee.
Buy my deep learning course at udemy. Just click on the course link in the show notes and get awesome discount on my deep learning course.
Subscribe to my publication and share it across so that more people can discover it.
Subscribe and share my podcast “SimpleAI” on google podcast or any other podcast player of your choice. Don’t forget to give it a 5 star.
Subscribe to my newsletter.