Devika Puri
Jan. 28
In his article, “Don’t forget people in the use of big data for development”, Blumenstock explores the promise of data science. He focuses on how data science can help to solve many humanitarian issues around the world, through the use of algorithms. For example, algorithms have found a correlation between thatched roofs of African houses and increased poverty level. Thus, satellite data of thatched roofs could allow governments to distribute aid accordingly to the people. Aside from aid, data science shows promise as it can provide lawmakers with information necessary to make policy changes.
After reviewing all the potential benefits of data science, Blumenstock does acknowledge the pitfalls of such a promising technology. First, he mentions that there may be some unantivipated effects. For example, one benefit of data science, discussed earlier, is to provide aid to those in poverty. However, sometimes, the solutions from big data may further empower the wealthy. Blumenstock also mentions that data collection methods can be flawed, and algorithms aren’t always accruate. This is especially true if people “game” the system. For example, if people with thatched roofs are receiving aid, then others may get a thatched roof too, just to receive benefits. Next, algorithms can be biased. For example, “poor people in emerging economies” tend to be the most underrrepresented in new sources of data. Even in my daily life, I can see how this is true. Data pulled from facebook or instagram does not reflect people of all the different socioeconomic backgrounds. Lastly, there is a lack of regulation in the field of data science. Although the US has rules about data privacy, many other countries do not. I think this issue is one of the most important, as it’s been very relevant in our daily lives. Privacy issues remain a debated controversy in our society, as seen with the recent facebook privacy scandal.
With the promises and pitfalls in mind, Blumenstock proposes how the field of data science could improve going forward. He mentions increased collaboration is needed by many indiividuals, like data scientists, government officials, and development experts, etc. Also, he states that old sources of data should not be completely abandoned for new sources of data, but instead encourage a side-by-side comparison of the two. Additionally, algorithms need to be customized to fit the local context of a place.
Overall, I generally agree with most of the statements from other classmates. Data science can be tricky, because its intentions are well. This wonderful field can help provide aid and generally improve welfare. However, the balancing act can be difficult because in order to make these changes, the people must give up some part of their personal life. There is definitly a huge trade-off between privacy and what can be accomplished. I do think transparency on both ends would help. For example, if people find out that their personal data is being taken (without prior knowledge), this creates distrust of the field of data science, and hinders its ability to do good. For this reason, more people should be knowledgable of the field in general - the promises and the pitfalls - to allow them to understand that some sacrifice is required to make powerful changes. Moreover, data scientists also have a responsibility to look beyond the data, and keep the big picture idea in mind. They must remember that behind the data, are real people living their lives - who could be impacted in many different ways.
Lastly to answer other questions, big data is large sets of data - everything froma simple bar code that was scanned to an instagram picture. Machine learning is the scientific study of algorithms that can access and utilize data. Human development is the science to understand how and why people change or remain the same over time.