Module 0 | Introduction to Business Intelligence

Hi all: 

Module 0 of this course introduces fundamental concepts of big data and business intelligence. I would highlight the shift of paradigm in data science caused by big data as the main take-away in this unit. This shift involves the "datafication " of the world (many more things are being tracked), richer and more dynamic information (production of data and information is growing rapidly), and the "N=All" notion (the entire population of data is available so sampling is not required). I think there may be cases where communities are not yet well represented in the data (so models may be biased), but this might change in the future as IT becomes more popular and pervasive in human life.

I found two articles, MIT’s Big Data Gets Personal and MGI’s Ten IT-enabled business trends, to be especially interesting, not only for the huge possibilities of business intelligence, big data and machine learning for human development, but also for the ethical questions that these technologies pose.

In a comment to one of our blogs, I have reflected on the potential of anticipatory systems and how they might affect fundamental rights. Thinking of how risk modeling is currently used in the insurance sector to determine premiums based on predictors such as age, gender or marital status, it made me wonder whether anticipated systems, in the absence of legal civil protections, might cause the segregation of communities based on predicted health or behavior. The potential of anticipatory systems being used against presumption of innocence does not seem so remote these days in a world with so many different ways of understanding the role of the government and the citizens. In the face of globalization, global policy solutions should be developed at the international level to address these potential issues, though technology seems to be evolving faster than information technology law.

Some of the most recent cases around this question have been summarized on Wired:

Sidney Fussell, An Algorithm That ‘Predicts’ Criminality Based on a Face Sparks a Furor, 06-24-2020

I also seem to see a risk in the reinforcement patterns that recommender and anticipation systems may encourage. Using a terminology that comes from heuristics and evolutionary computing, these systems seem to incentivize exploitation of known behavior before exploration of new knowledge (e.g. new experiences or perspectives on a question). The notion of augmented social reality also seems to be oriented to exploiting (social/cultural) patterns that are already known, rather than helping explore new ones. Classification of consumers based on a set of features might reinforce divided communities, as opposed to let interaction and exchange of knowledge happen between different communities. While exploitation of knowledge enables the optimization of a solution, exploration allows the search of new solutions as well as diversification of knowledge. I think one task should not be hyper-developed at the cost of the other.