Benchmarking explainable offensive language detection in Somali with human-annotated rationales.

Formal recommendations on alcohol use, called low-risk drinking guidelines, are a staple of alcohol public health efforts worldwide. Health agencies in the US, UK, Canada, Australia, and many other nations have issued low-risk drinking guidelines based on the health effects of consumption [–]. Although the scientific basis for guidelines varies by country, all rely on epidemiological evidence of the harmful health effects of drinking to establish limits based on either the absolute or relative risk of mortality or morbidity caused by alcohol use. Because both the relative and absolute risk of harm associated with drinking may vary based on the pattern of consumption, some guidelines specify both daily and weekly limits, although the usefulness of recommending both is a source of debate [–]. Many guidelines also recommend populations that should not drink at all, typically children and adolescents, people who are or may become pregnant, and those with an alcohol use disorder [–]. Despite the ubiquity of low-risk drinking guidelines, there is little evidence that they impact drinking behaviors. Early research on guidelines focused on public awareness of the guidelines themselves [, ]. Later studies assessed the effectiveness of guidelines but failed to find evidence that they reduced consumption []. Thus, the current evidence suggests that even when the public is aware of guidelines, they seldom reduce alcohol consumption at a population level. Nonetheless, guidelines still play an important role in alcohol public health messaging. Given their importance in setting consumption targets for alcohol policy and alcohol public health efforts, it is critical to develop low-risk drinking guidelines that are more effective in changing drinking behaviors. The process for developing low-risk drinking guidelines has been criticized, however, as lacking transparency, being dependent on subjective assessments of health risks that people will tolerate, or as paternalistic attempts to dictate the health risks that drinkers’ should accept [, ]. The morbidity and mortality focus of most guidelines ultimately result in guidelines that ignore, or at best indirectly incorporate, drinkers’ quality of life and preferences for health. Research consistently finds that low levels of alcohol use are associated with higher quality of life for the drinker [, ], yet no existing guidelines formally incorporate drinkers’ quality of life. Debates about revised guidelines in the US [] and in Canada [–] underscore the continuing tension between drinking limits that focus on increasing years lived rather than on the quality of life in those years or, at a more fundamental level, individuals’ subjective preference for health over other aspects of drinking behaviors, such as social interactions.