Collateral Damage in the Battle Over Truth
At a time when truth and facts are under attack and struggling for survival, the valiant defendants have pulled up the ramparts and reduced truth to empirical evidence and measurable, quantified, statistical proof. All other forms of proof or evidence are rejected and deemed suspect. Personal experiences are merely anecdotal, the unmeasurable is irrelevant, and success must scale a formula to be an evident success. There is a call to base all decisions on this reduced form of evidence — evidence-based medicine, evidence-based governance, and evidence-based budgeting.
The first casualty during struggles for survival is self-critique. The defendants of truth cannot risk questioning the boundaries that have been drawn. They must double down.
In this battle between truth and untruth, the collateral damage is the people whose truth does not meet the statistical thresholds, for whom the measures they need do not scale, who live far from the statistical average. It is the outliers and small minorities in any population.
If you plot the full range of the most critical needs of any population on a multi-variate scatterplot, the pattern is a normal distribution. It looks like a starburst (a “human starburst”). The starburst loosely adheres to the 80/20 pattern that Pareto observed. 80 percent of the needs take up the central 20 percent of the space and the remaining 20 percent are scattered in the 80 percent that is the peripheral space. The middle 80 percent are close together, the peripheral 20 percent are far apart meaning they are increasingly different from each other. The outer edge of the starburst is jagged.
Figure 1: “The human starburst”- needs of a population plotted in a multi-variate scatterplot showing the accuracy of any statistically determined truth relative to the position within the distribution.
Because of how truth has been reduced, any statistically determined, sanctioned truth is generally accurate for people with needs in the center, inaccurate as you move away from that core and wrong if your needs or life experiences happen to situate you at the jagged edges (Figure 1). The consequence of this is disparity — a widening chasm of disparate opportunity and prosperity.
This rigid reduction of truth has been mechanized, amplified, accelerated, and automated in the generation of AI that is already pervasively deployed to make some of the most critical decisions of our lives — what medical treatment you receive, whether you get a mortgage, what you are worth, whether you get an interview for a job, or are admitted to a university, whether you get to keep your kids, how long you must stay in jail, whether you are a security risk, whether the products you need will be marketed; as well as a million small decisions that add up to substantive cumulative harm. Outliers gum up this machinery. The people that are far from the average are consistently, accurately, and efficiently decided against.
The defenders of excluded minorities blame lack of representation in data and human biases corrupting the determinations. But even with full proportional representation and the removal of all bigoted human influence, the decision machines will still rule against the outliers and minorities.
In an ironic twist, emerging large language models and generative AI have shown that the same statistical reasoning is incredibly effective at creating believable “bullshit” and hallucinations. The sanctioned formula for certified truth is the most effective means of creating believable untruth.
Despite the present threat to truth, it is time to examine what we mean by truth, evidence, proof, best, impact, and success. Our current definitions and defended boundaries are powering us toward a monoculture and denying the truth and survival of all who find themselves at the jagged edge of the human starburst.
Cover image credit: Counting By Bhakti Ziek. — provided by the author, CC BY-SA 4.0
Bio:
Jutta Treviranus is the Director of the Inclusive Design Research Centre (IDRC) and professor in the faculty of Design at OCAD University in Toronto. Jutta established the IDRC in 1993 as the nexus of a growing global community that proactively works to ensure that our digitally transformed and globally connected society is designed inclusively. Dr. Treviranus also founded an innovative graduate program in inclusive design at OCAD University. Jutta is credited with developing an inclusive design methodology that has been adopted by large enterprise companies such as Microsoft, as well as public sector organizations internationally. In 2022 Jutta was recognized for her work in AI by Women in AI with the AI for Goodhttps://www.nytimes.com/2023/01/06/opinion/ezra-klein-podcast-gary-marcus.html — DEI AI Leader of the Year award.
Originally posted on OpenDataScience.com
Read more data science articles on OpenDataScience.com, including tutorials and guides from beginner to advanced levels! Subscribe to our weekly newsletter here and receive the latest news every Thursday. You can also get data science training on-demand wherever you are with our Ai+ Training platform. Subscribe to our fast-growing Medium Publication too, the ODSC Journal, and inquire about becoming a writer.
