Browse Articles

This study investigates skin tone bias in artificial intelligence models used for dermatological disease classification and evaluates a CycleGAN-based data augmentation approach to improve diagnostic performance on darker skin types. We generated synthetic dark-skinned images to enhance dataset diversity and compared model performance before and after augmentation. The results demonstrate that augmentation with synthetic dermatological images can help reduce disparities in diagnostic performance across skin tones, highlighting a practical strategy for improving fairness in dermatology AI systems.

Advertising platforms like Google Ads use AI to drive the algorithms used to maximize advertisers benefits. This study shows that AI does not adjust it strategy based on auction type and highlights the limitations of AI running without explicit guidance.

Various methods exist to mitigate bias in AI models, including "Constitutional AI," a technique which guides the AI to behave according to a list of rules and principles. Lo, Poosarla, Singhal, Li, Fu, and Mui investigate whether constitutional AI can reduce bias in AI outputs on political topics.

While remarkable in its ability to mirror human cognition, machine learning and its associated algorithms often require extensive data to prove effective in completing tasks. However, data is not always plentiful, with unpredictable events occurring throughout our daily lives that require flexibility by artificial intelligence utilized in technology such as personal assistants and self-driving vehicles. Driven by the need for AI to complete tasks without extensive training, the researchers in this article use fluid intelligence assessments to develop an algorithm capable of generalization and abstraction. By forgoing prioritization on skill-based training, this article demonstrates the potential of focusing on a more generalized cognitive ability for artificial intelligence, proving more flexible and thus human-like in solving unique tasks than skill-focused algorithms.

AI analysis of brain scans offers promise for helping doctors diagnose brain tumors. Haider and Drosis explore this field by developing machine learning models that classify brain scans as "cancer" or "non-cancer" diagnoses.

The authors employ computational protein design to identify a mini-protein with the potential to enhance binding of the tight junction protein, claudin-5, at the blood-blood barrier with therapeutic potential for neurodegenerative diseases.

The authors looked at different ways to build computational resources that would analyze shooting form for basketball players.

Using facial recognition as a use-case scenario, we attempt to identify sources of bias in a model developed using transfer learning. To achieve this task, we developed a model based on a pre-trained facial recognition model, and scrutinized the accuracy of the model’s image classification against factors such as age, gender, and race to observe whether or not the model performed better on some demographic groups than others. By identifying the bias and finding potential sources of bias, his work contributes a unique technical perspective from the view of a small scale developer to emerging discussions of accountability and transparency in AI.

This study evaluates the potential of natural language processing (NLP) models in an emotion-driven bibliotherapy framework to improve mental health challenges.

The study developed Loving Words, a free AI-powered poetry tutor designed to help writers improve their poetry and experience its therapeutic benefits. Two groups of participants wrote poems—one without assistance and one using Loving Words.