Algorithmic Bias, Systemic Inequality, and the Future of Higher Education Amidst COVID-19

Cathy O'Neil, an expert in algorithmic auditing, discusses the profound societal vulnerabilities exposed by the COVID-19 pandemic, arguing that the crisis is not merely a biological phenomenon but a stark revelation of deep-seated social system failures. She posits that issues like deepening inequality, mass incarceration, and systemic racism are causal factors in the epidemic's spread and severity, making certain populations disproportionately vulnerable. O'Neil illustrates this with examples from the US, where neighborhood health correlates with incarceration rates, and Singapore, where guest worker dormitories became hotspots despite initial contact tracing success. She emphasizes that true national resilience against future pandemics requires addressing these seemingly intractable social problems, highlighting a collective responsibility often overlooked in favor of individual blame.

A significant portion of the conversation critiques the use of "crime risk scores" in the justice system, particularly in decisions regarding prisoner release during the pandemic. O'Neil meticulously explains that these algorithms do not measure inherent criminality but rather predict who will be arrested, based on proxies for poverty, mental health status, addiction, and race. She argues that these scores, which are often static and racialized, perpetuate past biases and should be interpreted as indicators for social support and treatment rather than justification for prolonged incarceration. This distinction underscores a broader philosophical problem where algorithms, under the guise of scientific authority, are used to avoid difficult societal conversations about justice and equity.

Regarding higher education, O'Neil categorizes colleges into three tiers: elite, middle, and commuter schools, each facing distinct challenges. Elite institutions, with large endowments and older faculty, will likely survive by spending their assets to protect staff and adapt. Middle-tier liberal arts colleges, heavily reliant on tuition, face an existential threat and are likely to be decimated without significant changes. Commuter schools, focused more on education than a holistic experience, may transition more easily to online models. O'Neil anticipates a "game of chicken" between colleges and students/parents over tuition, likely leading to price negotiations as institutions struggle to justify full costs for potentially diminished experiences.

Broadly, the discussion extends to the "authority of the inscrutable" in quantitative analysis, where complex mathematical models are used to mask uncertainty and avoid accountability. Rob Johnson connects this to the economic concept of "ergodic stability," arguing that applying stable statistical models to inherently unstable systems driven by subjective human psychology is a form of "scientism" rather than true science. Both O'Neil and Johnson contend that this often serves to reinforce existing power structures and financial interests, rather than promoting public good. The pandemic, therefore, acts as a powerful lens, revealing the ethical compromises embedded within our analytical tools and institutional frameworks, urging a re-evaluation of how we define and address risk, justice, and societal well-being.

"inequality mass incarceration and racism are actually causing the epidemic"

"what we should be doing is addressing these things like inequality like mass incarceration like racism that seem impossibly difficult and they are very difficult but are actually 100 required in order to make ourselves healthier"

"we should see that as we should have taken care of those people better that is on us"

"you can't you can't just attribute it to the individual and so i just i think i think it's a dreadful escape hatch for people who don't want to address collective responsibility"

"their faculty is their largest asset and they really cannot afford to put their faculty at risk"

"algorithms propagate the past they look for patterns in the past and then they they predict that the pattern will repeat"

"what a crime risk score actually does a crime risk score predicts who will be arrested"

"this is not actually measuring a criminality the inherent criminality of a human"

"somebody with a higher score deserves more support like for example mental health support or addiction treatment"

"the authority of the inscrutable which we which we use as quants"