6.3 KiB
| title | chunk | source | category | tags | date_saved | instance |
|---|---|---|---|---|---|---|
| Algorithmic bias | 5/13 | https://en.wikipedia.org/wiki/Algorithmic_bias | reference | science, encyclopedia | 2026-05-05T16:31:03.393915+00:00 | kb-cron |
==== Racial bias ==== Racial bias refers to the tendency of machine learning models to produce outcomes that unfairly discriminate against or stereotype individuals based on race or ethnicity. This bias often stems from training data, which is shaped by humans' opinions, assumptions, and racial prejudices. These data lead AI systems to reproduce and amplify historical and systemic discrimination. For example, AI systems used in hiring, law enforcement, or healthcare may disproportionately disadvantage certain racial groups by reinforcing existing stereotypes or underrepresenting them in key areas. Such biases can manifest in ways like facial recognition systems misidentifying individuals of certain racial backgrounds or healthcare algorithms underestimating the medical needs of minority patients. Addressing racial bias requires careful examination of data, improved transparency in algorithmic processes, and efforts to ensure fairness throughout the AI development lifecycle. Empirical audits of deployed vision models also show race linked disparities in occupational labeling; for example, in Google Cloud Vision AI, women of color were the least likely to be identified as scientists, indicating compounding effects of race and gender in model outputs. Another clear indication of how racial biases are reproduced through technological advances is predictive policing. Predictive policing tools make assessments about who, when will future crimes be committed, and where any future crime may occur, based on location and personal data . This means specific areas and where there have been an uptick in crimes usually see more prediction of future crimes. For instance, Afghanistan nationals were largely restricted from purchasing ammonium fertilisers because it was discovered that most improvised explosive devices used against United States Of American soldiers contained sufficient amounts of nitrates which is a chief ingredient of ammonium fertilizers. This ban which was subsequently enforced with the use of artificial intelligence by U.S force saw even Afghan nationals whose sole means of livelihood or sustenance were through agriculture effectively denied a major agricultural input (fertilisers) because the AI used for enforcing this ban was primarily looking out for a blanket description of bearded Muslims or Afghan nationals . In China, most especially in the Muslim minority Xinjiang region, the use of AI to restrict Muslim minorities, otherwise known as ethnic Uyghurs goes far beyond banning specific materials . Here a system of automatic denial is largely used. Unlike the Afghan fertilizer ban, Chinese systems uses AI to define "suspicious behavior" and then automatically denies Uyghurs from being able to purchase household commodities such as kitchen knives , if they must, then there have to be serious set of protocols to be passed and this includes having a barcode of trustworthiness being etched on the knife with the barcode containing every ounce of personal data or identification of the purchasing Uyghur. By training artificial intelligence models to be able to predict or even be able of racial profiling, the system is unequivocally made to be racially biased.
==== Speciesist bias ==== Speciesist bias (also known as anthropocentric bias) refers to the tendency of large language models to systematically devalue or discriminate against non-human animals, often by prioritizing human interests or reinforcing the objectification of animals. This bias typically manifests as anthropocentrism, where the AI views animals primarily through their utility to humans (e.g., as food, tools, or pests) rather than as sentient beings with intrinsic value.
=== Technical ===
Technical bias emerges through limitations of a program, computational power, its design, or other constraint on the system. Such bias can also be a restraint of design, for example, a search engine that shows three results per screen can be understood to privilege the top three results slightly more than the next three, as in an airline price display. Another case is software that relies on randomness for fair distributions of results. If the random number generation mechanism is not truly random, it can introduce bias, for example, by skewing selections toward items at the end or beginning of a list. A decontextualized algorithm uses unrelated information to sort results, for example, a flight-pricing algorithm that sorts results by alphabetical order would be biased in favor of American Airlines over United Airlines. The opposite may also apply, in which results are evaluated in contexts different from which they are collected. Data may be collected without crucial external context: for example, when facial recognition software is used by surveillance cameras, but evaluated by remote staff in another country or region, or evaluated by non-human algorithms with no awareness of what takes place beyond the camera's field of vision. This could create an incomplete understanding of a crime scene, for example, potentially mistaking bystanders for those who commit the crime. Lastly, technical bias can be created by attempting to formalize decisions into concrete steps on the assumption that human behavior works in the same way. For example, software weighs data points to determine whether a defendant should accept a plea bargain, while ignoring the impact of emotion on a jury. Another unintended result of this form of bias was found in the plagiarism-detection software Turnitin, which compares student-written texts to information found online and returns a probability score that the student's work is copied. Because the software compares long strings of text, it is more likely to identify non-native speakers of English than native speakers, as the latter group might be better able to change individual words, break up strings of plagiarized text, or obscure copied passages through synonyms. Because it is easier for native speakers to evade detection as a result of the technical constraints of the software, this creates a scenario where Turnitin identifies foreign-speakers of English for plagiarism while allowing more native-speakers to evade detection.