PPECMC2021CovidCrusaders

 In her paper, “Are Algorithms Value-Free?”, Professor Johnson shows that all science, and consequently all computer algorithms, require assumptions or values to build knowledge. Scientists and algorithms use induction to draw conclusions from evidence. However, there is no correct way to link evidence to a certain conclusion. As Johnson says, “There are many possible bridges one might adopt to traverse the gap between evidence and theory, and there seem to be no a priori grounds for preferring some bridges over others.” (Johnson 5). Still, science has attempted to find some value-free standards to serve as objective, independent grounds for theory, such as accuracy, consistency, simplicity, and others. Failing to recognize the relativism within scientific theory has resulted in some significant missteps. For example, Johnson discusses the case of Ambien in which clinical trials failed to consider metabolic differences between men and women, resulting in the FDA suggesting women...

Johnson’s essay asserts that inductive reasoning entails making certain assumptions, or canons. Deciding which of these canons to adopt-- epistemic vs. non-epistemic, for example-- entails being allied to certain values over others. Further, inductive risk necessitates adopting a “threshold for confidence [that] can only be established by appeal to ethical values” (Johnson 13). These examples serve as objections to the value-free ideal, allowing Johnson to ultimately contend that “the extent to which many of the real-world applications of machine learning today are useful to the extent that they are undeniably value-laden” (Johnson 20).  I’m especially intrigued by the AIPIE model that Johnson outlines in Section 4, and how the interaction between humans and algorithms may lead to more or less biased results. Johnson outlines the five-step AIPIE model as “Assessment, Interpretation of results, Plan based on the information gathered, Implement the plan, and Evaluate the results ...

  Johnson’s discussion of value laden algorithms and societal norms in “Are Algorithms Value-Free?” challenges the notion of oppression. In the early part of her argument, she establishes that “ canons of inference are necessary means of overcoming underdetermination” (5). Some principles need to be taken for granted in the world in order to accomplish the aims of science. However, other epistemic values, such as the ones that were established under the basis of gender oppression, need to be updated. I would like to look into and discuss this idea, and how it is very difficult to accept epistemic values given the ever changing nature of societal norms. Johnson introduces the example of a sleep medicine that was prescribed to fit the needs of men, but not of women. Not only that, but the issue took two decades to fix. Johnson points out that “In the case of novelty, the claim is that feminists adopt this virtue on the socio-political basis of aiming to depart ...

  In her paper “Are Algorithms Value-Free? Feminist Theoretical Virtues in Machine Learning,” Gabbrielle M. Johnson argues that machine learning is necessarily value-laden because it operates via inductive reasoning, which is value-laden. As such, the presupposed ideal for value-free decision making should not be the standard we aim to achieve for machine learning to make objective decisions. She argues that all inductive reasoning is value-laden because in order to draw conclusions, they necessitate “canons of inductive inference,” which are non-evidentiary ways to limit the hypothesis to prevent underdetermination. A hypothesis can never be proven 100 percent true, so researchers need to make a value judgment on how likely a hypothesis is to be true such that he could conclude that the evidence is “sufficient” enough.  Johnson raises a counter-argument against her case. When researchers test hypotheses, they are most often “assign probabilities to hypotheses with respect to ...

  Towards the end of her paper “Are Algorithms Value-Free?” Gabbrielle Johnson presents the debate on presenting degrees of confidence or probabilities instead of accepting or rejecting hypotheses. Johnson clearly explains why this doesn’t offer a solution but merely “pushes the problem back a level” (pg 17) she offers a further counterargument to that, which she refutes with questioning if scientists can be idealized Bayesian agents. While the discussion of Bayesian agents extends past the topic of the paper, this debate with confidence/probabilities versus accepting/rejecting a hypothesis reminds me of earlier discussions of coercion in decision making. One could refute the argument that proposing degrees of confidence and probability are not the same as the acceptance of a hypothesis by saying that providing a strong degree of confidence or probability is so difficult to refute, that it is no longer a real choice for lets say the judge who is relying on this information to make ...

        While reading Professor Johnson’s paper, I noticed many interesting connections and counterpoints to what has been drilled into my biological science education. I think it might be useful to point some of these out and apply them to the sphere of medicine with some questions that the class might discuss. Particularly in the section that connects scientific inquiry to machine learning, I was a bit reluctant to agree with the idea that “ like scientific inductions, machine learning programs use evidence (or known data) to form predictions (or generalizations to new ideas)” (2). While I somewhat agree with this statement, it is essential to discuss the difference between machine learning through observation and scientific study through experimentation. Observational studies, unlike experiments, can only be used to draw correlations. Experiments, on the other hand, can provide causal explanations. Writing this, I already see an interesting application of Joh...

   In her article, “Are Algorithms Value-Free?” Gabrielle Johnson argues that algorithmic decision-making is far less objective than is generally assumed. She describes “Dragnet objectivity” in the context of machine learning, which is the (incorrect) assumption that algorithmic decision-making is objective because it learns from raw data and is thus devoid of the flaws inherent to human decision-making, such as “personal speculation or emotional interest” (3). Unfortunately, because relying on raw data alone leads to underdetermination, inductive conclusions cannot be drawn on “just the facts.” As Johnson explains, underdetermination can only be overcome with the creation of a set of assumptions, which she dubs “canons of inductive inference” (5). Throughout her article, Johnson draws compelling parallels between scientific inquiry and machine learning. Another parallel that may be pertinent involves statutory interpretation. Johnson’s description of the role of canons ev...