Reflection #2
Personal Understanding of Computations Impact on Social Values
Taking social media as an example, I can’t escape the feeling that algorithms have at most times exposed or accentuated our personal biases or improper social values through our digital behaviour, subconscious as it may be at most times.
And it reflects in the recommendations to content, ads, opinions and thoughts that we are exposed to. The information silos that people get warped into has been a telling description of the political landscape in many major countries across the world. It has enabled people to exist in their own realities of information and understanding with an almost alarming lack of empathy among humans for one another, as a result of it all.
Therefore, the studies on misinformation, disinformation and other interventions for social change were a refreshing introduction to the cause for me. And the paper shed light on yet more such interventions that AI could be at the forefront of that piqued my interest.
Overview of the Paper
The paper highlighted the use of computation as interventions to problems in society — premised around technical and non-technical systems. I appreciate the authors’ acknowledgement that the interventions they offer were modest and solutions far from optimal, with potential hazards of their own.
It speaks about being a part of an intentional and meaningful advancement towards social change and using computational systems as a means to help that cause.
The paper underlines three different approaches through which computing can be used to address social problems:
- Computing As Diagnostic: In short, this talks about identifying social problems and how they manifest in our technical systems.
- Computing As Formalizer: This talks about using computing to disambiguate vagueness and provide more detail and context to how social problems are understood.
- Computing As Rebuttal: This talks about using computational systems to underscore the limitations of interventions that are premised on them.
As an approach, I was able to relate to, understand and appreciate Computing being used to Diagnose problems and being used as a rebuttal to other technical interventions.
One of my favorite lines from the description of this practise was:
“Other recent work has sought to advance the diagnostic role that computing can play by providing a principled approach to inter- rogating the machine learning pipeline.These diagnostic efforts do not present themselves as solutions, but rather as tools to rigorously document practices. Thus, when compared with other computing interventions that aim directly at incremental improvements, they are less vulnerable to becoming a substitute for broader change.”
I particularly appreciated it because that is a mindset that needs to be embraced more widely. The idea of documenting practices and critically examining a system in question doesn’t exactly fit into the idea of incremental improvement for most major stakeholders in the tech industry.
Likewise, with the Rebuttal approach, I liked the idea of using tech systems to check on the limits of other systems designed to be interventions. With fact-checkers and advent of social media misinformation tools, there is very possibly a case for a computational system in the future that checks the limits, legitimacy of some of these tools in question. This in my mind feels like it can help develop a sense of accountability to the design of computational systems to solve problems.
I wasn’t entirely sure or convinced about Computing as Formalizer as a method. When there is as much vagueness in the human understanding, practise of implementation of vague rules and regulations, won’t computational systems only accentuate that even more? Even if they can be used to identify and address that vagueness, the process feels similar to Computing to Diagnose? I fully acknowledge that this might also be a limitation in my understanding of it as a method.
