Causal inconsistency arises when the underlying causal graphs captured by generative models like \textit{Normalizing Flows} (NFs) are inconsistent with those specified in causal models like \textit{Struct Causal Models} (SCMs). This inconsistency can cause unwanted issues including the unfairness problem. Prior works to achieve causal consistency inevitably compromise the expressiveness of their models by disallowing hidden layers. In this work, we introduce a new approach: \textbf{C}ausally \textbf{C}onsistent \textbf{N}ormalizing \textbf{F}low (CCNF). To the best of our knowledge, CCNF is the first causally consistent generative model that can approximate any distribution with multiple layers. CCNF relies on two novel constructs: a sequential representation of SCMs and partial causal transformations. These constructs allow CCNF to inherently maintain causal consistency without sacrificing expressiveness. CCNF can handle all forms of causal inference tasks, including interventions and counterfactuals. Through experiments, we show that CCNF outperforms current approaches in causal inference. We also empirically validate the practical utility of CCNF by applying it to real-world datasets and show how CCNF addresses challenges like unfairness effectively.

Bias detection and mitigation is an active area of research in machine learning. This work extends previous research done by the authors to provide a rigorous and more complete analysis of the bias found in AI predictive models. Admissions data spanning six years was used to create an AI model to determine whether a given student would be directly admitted into the School of Science under various scenarios at a large urban research university. During this time, submission of standardized test scores as part of an application became optional which led to interesting questions about the impact of standardized test scores on admission decisions. We developed and analyzed AI models to understand which variables are important in admissions decisions, and how the decision to exclude test scores affects the demographics of the students who are admitted. We then evaluated the predictive models to detect and analyze biases these models may carry with respect to three variables chosen to represent sensitive populations: gender, race, and whether a student was the first in his or her family to attend college. We also extended our analysis to show that the biases detected were persistent. Finally, we included several fairness metrics in our analysis and discussed the uses and limitations of these metrics.

This article addresses the challenge of validating the admission committee's decisions for undergraduate admissions. In recent years, the traditional review process has struggled to handle the overwhelmingly large amount of applicants' data. Moreover, this traditional assessment often leads to human bias, which might result in discrimination among applicants. Although classical machine learning-based approaches exist that aim to verify the quantitative assessment made by the application reviewers, these methods lack scalability and suffer from performance issues when a large volume of data is in place. In this context, we propose deep learning-based classifiers, namely Feed-Forward and Input Convex neural networks, which overcome the challenges faced by the existing methods. Furthermore, we give additional insights into our model by incorporating an interpretability module, namely LIME. Our training and test datasets comprise applicants' data with a wide range of variables and information. Our models achieve higher accuracy compared to the best-performing traditional machine learning-based approach by a considerable margin of 3.03\%. Additionally, we show the sensitivity of different features and their relative impacts on the overall admission decision using the LIME technique.

University admission at many highly selective institutions uses a holistic review process, where all aspects of the application, including protected attributes (e.g., race, gender), grades, essays, and recommendation letters are considered, to compose an excellent and diverse class. In this study, we empirically evaluate how influential protected attributes are for predicting admission decisions using a machine learning (ML) model, and in how far textual information (e.g., personal essay, teacher recommendation) may substitute for the loss of protected attributes in the model. Using data from 14,915 applicants to an undergraduate admission office at a selective U.S. institution in the 2022-2023 cycle, we find that the exclusion of protected attributes from the ML model leads to substantially reduced admission-prediction performance. The inclusion of textual information via both a TF-IDF representation and a Latent Dirichlet allocation (LDA) model partially restores model performance, but does not appear to provide a full substitute for admitting a similarly diverse class. In particular, while the text helps with gender diversity, the proportion of URM applicants is severely impacted by the exclusion of protected attributes, and the inclusion of new attributes generated from the textual information does not recover this performance loss.

In December, the University of Texas at Austin's computer science department announced that it would stop using a machine-learning system to evaluate applicants for its Ph.D. program due to concerns that encoded bias may exacerbate existing inequities in the program and in the field in general. This move toward more inclusive admissions practices is a rare (and welcome) exception to a worrying trend in education: Colleges, standardized test providers, consulting companies, and other educational service providers are increasingly adopting predatory, discriminatory, and outright exclusionary student data practices. Student data has long been used as a college recruiting and admissions tool. In 1972, College Board, the company that owns the PSAT, the SAT, and the AP Exams, created its Student Search Service and began licensing student names and data profiles to colleges (hence the college catalogs that fill the mail boxes of high school students who have taken the exams). Today, College Board licenses millions of student data profiles every year for 47 cents per examinee.

Given a huge set of applicants, how should a firm allocate sequential resume screenings, phone interviews, and in-person site visits? In a tiered interview process, later stages (e.g., in-person visits) are more informative, but also more expensive than earlier stages (e.g., resume screenings). Using accepted hiring models and the concept of structured interviews, a best practice in human resources, we cast tiered hiring as a combinatorial pure exploration (CPE) problem in the stochastic multi-armed bandit setting. The goal is to select a subset of arms (in our case, applicants) with some combinatorial structure. We present new algorithms in both the probably approximately correct (PAC) and fixed-budget settings that select a near-optimal cohort with provable guarantees. We show on real data from one of the largest USbased computer science graduate programs that our algorithms make better hiring decisions or use less budget than the status quo. '... nothing we do is more important than hiring and developing people. At the end of the day, you bet on people, not on strategies." - Lawrence Bossidy, The CEO as Coach (1995)

Joy Buolamwini is a graduate researcher at the MIT Media Lab and founder of the Algorithmic Justice League – an organisation that aims to challenge the biases in decision-making software. She grew up in Mississippi, gained a Rhodes scholarship, and she is also a Fulbright fellow, an Astronaut scholar and a Google Anita Borg scholar. Earlier this year she won a $50,000 scholarship funded by the makers of the film Hidden Figures for her work fighting coded discrimination. How did you become interested in that area? When I was a computer science undergraduate I was working on social robotics – the robots use computer vision to detect the humans they socialise with.

This article describes GRADE, a statistical machine learning system developed to support the work of the graduate admissions committee at the University of Texas at Austin Department of Computer Science (UTCS). In recent years, the number of applications to the UTCS PhD program has become too large to manage with a traditional review process. GRADE uses historical admissions data to predict how likely the committee is to admit each new applicant. It reports each prediction as a score similar to those used by human reviewers, and accompanies each by an explanation of what applicant features most influenced its prediction. GRADE makes the review process more efficient by enabling reviewers to spend most of their time on applicants near the decision boundary and by focusing their attention on parts of each applicant’s file that matter the most. An evaluation over two seasons of PhD admissions indicates that the system leads to dramatic time savings, reducing the total time spent on reviews by at least 74 percent.