Submission:
Speakers are required to submit abstracts on OpenReview.
https://openreview.net/group?id=brown.edu/Brown_University/2026/MLSJ#tab-your-consoles
Submission Start Date: 21 May 2026, 8:00 am EST
Submission Deadline: 30 Jun 2026, 11:59 pm EST
*Important note: In the submission title, please indicate Oral, Workshop, or Lightning. Else, abstracts will be reviewed based on paper length and desk-rejected if beyond length requirements.
The Call:
The development of many Machine Learning (ML) tools rarely considers the contexts in which decision-making would be used as an assistive partner or concluding party. This lack of consideration has led to damaging outcomes and even irreversible consequences for communities that are already marginalized. There have been several attempts to understand how we can correct this lack of presupposition through penalizing ML algorithms in the training phases and post-development using “fairness” as a framework, oftentimes through fairness metrics. These fairness metrics are used to quantify the extent to which outcomes and model trajectories diverge from fair outcomes. Naturally, this requires us to make some assumptions about the underlying data—many of which are unrealistic in real-world contexts. For instance, demographic parity is a widely used metric that suggests different demographic groups should experience equal rates of positive outcomes (Agarwal et al., 2019). However, this overlooks the fact that marginalized communities may appear less qualified due to historic exclusion from key social, educational, or professional spaces, despite being equally or even more qualified and capable. Further, these definitions fail to account for the severe data skews affecting marginalized populations, which distort both model inputs and outcomes (Pozzolo et al., 2013). These missteps in data distributions and collections, along with poor feature selection, lead to disparate treatment and disparate impact of marginalized individuals. A much better aim, therefore, is to consider not what is simply equal or equitable, but what would remove barriers in ways that are accurate and equitable, a concept known as justice. Justice also involves understanding what a community needs rather than meeting a perceived need (Benjamin et al., 2022).
The goal of this conference is to highlight the importance of justice in ML design and implementation. Particularly, we focus on the development and deployment of ML tools that promote a society where individuals receive what they need to overcome barriers to their participation as active and free members of society. Thus, we are seeking speakers who can make extrapolations and particular linkages to life-critical and high-impact contexts in which justice is considered; at this conference, we focus on Education, Healthcare, and Public Policy, but other contexts are welcome.
Format:
*Important note: In the submission title, please indicate Oral, Workshop, or Lightning. Else, abstracts will be reviewed based on paper length or desk-rejected if beyond capacity.
We seek speakers across a broad range of disciplines, regardless of experience level, to submit oral talk abstracts (1 page) and community-oriented workshop papers (2 pages). Students may submit lightning talk abstracts (max 1 page). We recommend the 2-column style of the AAAI-2025 Author Kit, but do not require any particular formatting. All submissions are double-blind and non-archival. Thus, authors may submit previous works, works in progress, or works currently under review. If the work is currently under review, please see the venue's rules on anonymization, double-submissions, presentations, etc...