← All Posts /

Ethics in Data Science

With the ubiquitous use of data science in nearly every aspect of life, it is crucial for the minds behind this powerful instrument to be mindful of the influence that their models are having on every individual who comes in touch with their models and algorithms.

“With Great power comes greater responsibility,” as the saying goes. In today’s society, data is becoming as valuable as gold to large corporations. Where does data power come from, and what are the responsibilities for data scientists, in an era when all the big organisations compete to collect and preserve data and gather even more from other sources to get better insights into where data power comes from and what are the responsibilities for data scientists?

Let’s start with a discussion about power. The economist published the following headline in May 2017: “The world’s most precious resource is no longer oil, but data.” The comparison of “data as the new oil” dates back to 2006, when Clive Humbey of Tesco in the United Kingdom stated that data is the new oil. It’s precious, but it can’t be utilised unless it’s polished, which is where you come in. Data is a really valuable resource, but it’s up to you to extract the value. Data has a lot of potential, and it’s up to you to realise it. As a data scientist, you hold a strong and privileged position, and your abilities are in great demand. Most individuals do not have access to education in these areas for a number of reasons, including the difficulty of teaching statistics and programming, as well as a scarcity of teachers.

To do it, getting exposed to and encouraged to pursue skills in data science is rare, especially among certain groups such as women or rural populations; a master’s degree in data science is expensive and thus prohibitive for many, and we lack widespread statistical literacy to make the path to becoming a data scientist easier. Quantitative methods are frequently valued in Western society over observational reports, lived experience, or even rigorous qualitative analyses. This has historical roots in misogyny and colonialism when only a small percentage of people were thought to be intelligent enough to pursue logical subjects like math and statistics. Of course, science is important to society, and quantitative methods may teach us a lot, but they are overrated, even though many statistical conclusions are demonstrably false.

Remember the big data myths of Boyd and Crawford? That way, you can respect authority and power just because you are a data scientist. He has a variety of talents, from programming to visualization to communication. Let’s talk about responsibilities. When you think about data science, you might think of a business model that optimizes advertising revenue. While enterprise data science is used in all possible areas, from marketing to medicine, transportation to waste management, data scientists may find it close to implementing real-world work. Models and analysis ultimately affect real life. This is a decision you make because it’s convenient and tidy, or because you don’t completely know what the model is doing. Yes, this will happen to you and will do a lot of damage.

As a data scientist, it is our responsibility to think critically about algorithm design and, in the event of ambiguity, to communicate the algorithm’s role to the general public. This is known as context-sensitive data science, and it attempts to assist rather than replace the design of the job. Because the model and analysis are handy and neat, or because we don’t totally understand what the model is doing, they will eventually influence the judgments we make. Yes, it will be really harmful to you. Let’s look at an example of a not-so-obvious error in an algorithm that snowballs into unjustly affecting a large number of people risk assessment scores for determining whether or not someone is likely to commit a crime.This isn’t science fiction this is real life risk assessment scores which will be used in the criminal justice system today because people tend to see numbers and statistics as more objective.

Our job as a data scientist is to think critically about algorithm design and teach the general public how algorithms work. If in doubt, ask model stakeholders to interfere with the design, be friendly, curious, and remain critical.

Recommended » Benford's Law