To recognize and mitigate the harms of generative AI systems, it is crucial to consider whether and how different societal groups are represented by these systems. A critical gap emerges when naively measuring or improving is represented, as this does not consider people are represented. In this work, we develop GAS(P), an evaluation methodology for surfacing distribution-level group representational biases in generated text, tackling the setting where groups are unprompted (i.e., groups are not specified in the input to generative systems). We apply this novel methodology to investigate gendered representations in occupations across state-of-the-art large language models. We show that, even though the gender distribution when models are prompted to generate biographies leads to a large representation of women, even representational biases persist in how different genders are represented. Our evaluation methodology reveals that there are statistically significant distribution-level differences in the word choice used to describe biographies and personas of different genders across occupations, and we show that many of these differences are associated with representational harms and stereotypes. Our empirical findings caution that naively increasing (unprompted) representation may inadvertently proliferate representational biases, and our proposed evaluation methodology enables systematic and rigorous measurement of the problem.

Large Audio-Language Models (LALMs) are increasingly deployed in real-world applications, yet their robustness against malicious audio injection attacks remains underexplored. This study systematically evaluates five leading LALMs across four attack scenarios: Audio Interference Attack, Instruction Following Attack, Context Injection Attack, and Judgment Hijacking Attack. Using metrics like Defense Success Rate, Context Robustness Score, and Judgment Robustness Index, their vulnerabilities and resilience were quantitatively assessed. Experimental results reveal significant performance disparities among models; no single model consistently outperforms others across all attack types. The position of malicious content critically influences attack effectiveness, particularly when placed at the beginning of sequences. A negative correlation between instruction-following capability and robustness suggests models adhering strictly to instructions may be more susceptible, contrasting with greater resistance by safety-aligned models. Additionally, system prompts show mixed effectiveness, indicating the need for tailored strategies. This work introduces a benchmark framework and highlights the importance of integrating robustness into training pipelines. Findings emphasize developing multi-modal defenses and architectural designs that decouple capability from susceptibility for secure LALMs deployment.

Artificial intelligence can differentiate between the brain patterns of boys and girls aged 9 to 10 years old according to their sex, and possibly their gender – but not everyone is convinced by the accuracy of the results. The prevalence of conditions such as pain, headache and heart disease differs between the sexes, but we know little about the neurological variations here or between genders, particularly among children. To learn more, Elvisha Dhamala at the Feinstein Institutes for Medical Research in New York and her colleagues analysed thousands of sets of magnetic resonance imaging (MRI) data from more than 4700 children, with a roughly even split between the sexes. The children were all aged 9 to 10 and are participating in the Adolescent Brain Cognitive Development project. Sex was defined according to someone's "anatomy, physiology, genetics and/or hormones at birth".

Large Language Models (LLMs) especially ChatGPT have produced impressive results in various areas, but their potential human-like psychology is still largely unexplored. Existing works study the virtual personalities of LLMs but rarely explore the possibility of analyzing human personalities via LLMs. This paper presents a generic evaluation framework for LLMs to assess human personalities based on Myers Briggs Type Indicator (MBTI) tests. Specifically, we first devise unbiased prompts by randomly permuting options in MBTI questions and adopt the average testing result to encourage more impartial answer generation. Then, we propose to replace the subject in question statements to enable flexible queries and assessments on different subjects from LLMs. Finally, we re-formulate the question instructions in a manner of correctness evaluation to facilitate LLMs to generate clearer responses. The proposed framework enables LLMs to flexibly assess personalities of different groups of people. We further propose three evaluation metrics to measure the consistency, robustness, and fairness of assessment results from state-of-the-art LLMs including ChatGPT and GPT-4. Our experiments reveal ChatGPT's ability to assess human personalities, and the average results demonstrate that it can achieve more consistent and fairer assessments in spite of lower robustness against prompt biases compared with InstructGPT.

The proliferation of personalized recommendation technologies has raised concerns about discrepancies in their recommendation performance across different genders, age groups, and racial or ethnic populations. This varying degree of performance could impact users' trust in the system and may pose legal and ethical issues in domains where fairness and equity are critical concerns, like job recommendation. In this paper, we investigate several potential factors that could be associated with discriminatory performance of a recommendation algorithm for women versus men. We specifically study several characteristics of user profiles and analyze their possible associations with disparate behavior of the system towards different genders. These characteristics include the anomaly in rating behavior, the entropy of users' profiles, and the users' profile size.

While using a speaker verification (SV) based system in a commercial application, it is important that customers have an inclusive experience irrespective of their gender, age, or ethnicity. In this paper, we analyze the impact of gender and age on SV and find that for a desired common False Acceptance Rate (FAR) across different gender and age groups, the False Rejection Rate (FRR) is different for different gender and age groups. To optimize FRR for all users for a desired FAR, we propose a context (e.g. gender, age) adaptive thresholding framework for SV. The context can be available as prior information for many practical applications. We also propose a concatenated gender/age detection model to algorithmically derive the context in absence of such prior information. We experimentally show that our context-adaptive thresholding method is effective in building a more efficient inclusive SV system. Specifically, we show that we can reduce FRR for specific gender for a desired FAR on the voxceleb1 test set by using gender-specific thresholds. Similar analysis on OGI kids' speech corpus shows that by using an age-specific threshold, we can significantly reduce FRR for certain age groups for desired FAR.

Given messages The elimination of discrimination is an important with the same content for different genders, issue that our society is facing. Learning from dialogue models could produce biased responses, human behaviors, machine learning algorithms which have been measured in terms of their politeness have been proven to inherit the prejudices from and sentiment, as well as the existence of humans (Mehrabi et al., 2019). A variety of AI applications biased words (Liu et al., 2019a). Table 1 shows one have demonstrated common prejudices example from a generative dialogue model trained towards particular groups of people (Rodger and on the Twitter dialogue corpus. When we change Pendharkar, 2004; Howard and Borenstein, 2018; the words in the message from "he" to "she", the responses Rose, 2010; Yao and Huang, 2017; Tolan et al., produced by the dialogue model are quite 2019). It is evident from recent research that different. In particular, the dialogue model generates learning-based dialogue systems also suffer from responses with negative sentiments for females.

Facebook is reportedly aiming to cash in on the £11billion lonely hearts industry with a dating app to rival the likes of Tinder. The social media giant is internally testing the service with Facebook employees, according to a Twitter user who appeared to leak screenshots of the app on Twitter. The screenshots show that'Facebook Dating' would offer users a choice of five genders: woman, trans woman, man, trans man and non-binary people. The app also appears to allow users to prevent their current Facebook friends from seeing their profile or only match with people who they have mutual friends with. Dating apps like Bumble already ask users to sign in using their Facebook accounts in a bid to spot mutual friends – and avoid making embarrassing connections with family members or close friends.