AI in the Library

[...]chatbots are prone to proposing incremental solutions to environmental challenges that draw heavily on past experience and avoid more radical changes to existing economic, social, and political systems. We also find that chatbots are reluctant to assign accountability to investors and avoid associating environmental challenges with broader social justice issues.

Recent studies demonstrate that machine learning algorithms can discriminate based on classes like race and gender. In this work, we present an approach to evaluate bias present in automated facial analysis algorithms and datasets with respect to phenotypic subgroups...We evaluate 3 commercial gender classification systems using our dataset and show that darker-skinned females are the most misclassified group (with error rates of up to 34.7%). The maximum error rate for lighter-skinned males is 0.8%.

The primary objective [of this study] is to evaluate implicit bias among GenAI across medical specialties. To evaluate for potential implicit bias, 100 photographs for each medical specialty were generated using the GenAI platform DALL-E2. GenAI created a sample population of physicians that underrepresented women when compared to both the resident and active physician workforce.