FazelpourDanks2021AlgorithmicBias

S. Fazelpour and D. Danks

Algorithmic bias: senses, sources, solutions

Bibliographic information

Fazelpour, Sina & Danks, David. (2021). Algorithmic bias: Senses, sources, solutions.
Philosophy Compass. 16. 10.1111/phc3.12760.

Commentary

The first thing that stood out to me was that the structure is proper. The article starts off with a disclaimer that related topics like trust and transparency will be omitted in favor of being able to discuss algorithmic bias adequately. Moreover, I liked the fact that they frequently referred to the student success example, making the article more coherent.
As a last note on the structure, the sources of bias were described in a logical order in which they could appear in the process of building and using an algorithm.
In terms of the content, they are thorough in describing all possible biases, and they
make sure that those biases are put in a nuanced light, in my opinion. In terms of
weaknesses, however, the paper provides different popular debiasing strategies but
then proceeds to provide arguments why they are difficult to use and how they might shift the bias instead of eliminating it. They suggest a way forward by focusing on the social epistemology and broadening their scope. In my opinion, these suggestions are too vague and abstract to function as an answer. The future directions the authors
describe could use more elaboration.

“Not all statistically (or legally) biased behaviors are ethically or morally problematic, while not all statistically fair or unbiased predictions are ethically or morally acceptable.”

This quote goes to show how contextual bias is. Bias is not inherently harmful. However,
the consequences of using an algorithm that is biased in a certain way could be
harmful.

“Different values in political settings sometimes imply the same policy. In contrast, different values almost always imply different objective functions for learning a model and so almost never result in the same algorithm.”

This begs the question of whether algorithms should be used for policymaking. The text
states that ‘even small differences in values lead to underdetermination of algorithms
(and possible biases)’. This is a problem, since we live in a value-pluralistic society. Is it
worth using algorithms in order to gain efficiency at the expense of fair treatment?

“These trade-offs are all heightened when algorithms are used repeatedly for multiple decisions, and so we may have to decide whether to allow some short-run ethical harms in order to gain additional knowledge that can enable long-term reduction in ethical harms.”

They make an interesting point here. Given that current research focuses on minimizing
immediate errors and biases, it is possible that a solution that will prove to be a good
solution later will be missed. I had not considered this argumentation myself

Commentary

The text gives a good overview of what algorithmic bias is and how it can arise in various ways. However there is not really a clear solution. The solutions section however, mostly describes why certain things don’t work. The solution they do give (fix the underlying real-world problem) seems a bit shortsighted. It also does not go in depth on the debate of what counts as a bias/discrimination. Overall, they don’t elaborate a lot on the positive side of using algorithms whilst still trying to claim that we should not stop using algorithms altogether.

Normative standards

• Page 2:

An algorithm can be morally, statistically, or socially biased (or other), depending on the normative standard used.

Comment:

I think this quote is nice especially due to the second part of the sentence. I wished the authors would have elaborated more on this. But bias is dependent on the normative standard used and that is an important thing to be aware of.

Fixing real-world problems

• Page 9:

In particular, when biases are deeply entrenched in a particular organizational or societal setting, then the proper response should often be to try to fix the underlying real‐world problem (Antony, 2016; Barabas et al., 2018; Mayson, 2018), rather than turning to the technical methods of fair ML.

Comment:

I think this is a bit short-sighted. Although fixing the underlying real-world problem is very important, algorithms are going to be created nonetheless. This does not give a solution on how to navigate that. Whilst I do agree that technical methods of fair ML are not the only solution, they do give a good measurable baseline of how fair ML should look like. Although critical reasoning is still needed to employ these methods, further assessment is probably also necessary in other parts of the process.

Value-laden vs. Value-free

Page 3: Algorithms are not objective, but rather embody the value‐laden view that some performance is better or more important than others

Comment:

This is a very extensive debate captured in one sentence. I think I very much agree with this. Algorithms cannot be truly objective when both the creation and the use of it is being done by subjective beings.