Lower performance for some languages and social groups

Taxonomy of Risks posed by Language Models

Weidinger et al. (2022)

Source DOI

Sub-category

Risk Domain

1Discrimination & Toxicity

1.3Unequal performance across groups

Accuracy and effectiveness of AI decisions and actions are dependent on group membership, where decisions in AI system design and biased training data lead to unequal outcomes, reduced benefits, increased effort, and alienation of users.

"LMs are typically trained in few languages, and perform less well in other languages [95, 162]. In part, this is due to unavailability of training data: there are many widely spoken languages for which no systematic efforts have been made to create labelled training datasets, such as Javanese which is spoken by more than 80 million people [95]. Training data is particularly missing for languages that are spoken by groups who are multilingual and can use a technology in English, or for languages spoken by groups who are not the primary target demographic for new technologies."(p. 217)

Entity— Who or what caused the harm

Human

Due to a decision or action made by humans

AI system

Due to a decision or action made by an AI system

Other

Due to some other reason or is ambiguous

Intent— Whether the harm was intentional or accidental

Intentional

Due to an expected outcome from pursuing a goal

Unintentional

Due to an unexpected outcome from pursuing a goal

Other

Without clearly specifying the intentionality

Timing— Whether the risk is pre- or post-deployment

Pre-deployment

Occurring before the AI is deployed

Post-deployment

Occurring after the AI model has been trained and deployed

Other

Without a clearly specified time of occurrence

Supporting Evidence (2)

"Training data can also be lacking when relatively little digitised text is available in a language, e.g. Seychellois Creole [95]. Disparate performance can also occur based on slang, dialect, sociolect, and other aspects that vary within a single language [23]. One reason for this is the underrepresentation of certain groups and languages in training corpora, which often disproportionately affects communities who are marginalised, excluded, or less fre- quently recorded, also referred to as the ”undersampled majority” [150]"(p. 217)

"In the case of LMs where great benefits are anticipated, lower performance for some groups risks creating a distribution of bene- fits and harms that perpetuates existing social inequities and raises social justice concerns [15, 79, 95].]"(p. 217)

Part of Risk area 1: Discrimination, Hate speech and Exclusion