Workshop on Datasets and Evaluators of AI Safety
Key information (Call for Papers)
Workshop Title: Datasets and Evaluators for AI Safety
Date: March 3rd 2025
Notification for Authors: December 14th 2024
Location: AAAI 2024 (Philadelphia, Pennsylvania, USA)
Duration: Full-day, approximately 6 hours of content, excluding breaks
Background
Advanced AI systems have the potential to drive economic growth and productivity, boost health and well-being, improve public services, and increase security. However, AI models can also cause societal harms and can be misused. This workshop focuses on evaluating the safety of AI models and in particular on LLMs. We are especially interested in work on improving datasets and benchmarks, as well as devising methods for evaluating the safety of AI models through the development of evaluators.
The goals of this full-day workshop, organised in collaboration with Kaggle, King's College London and the Open Data Institute are to bring together academic and industrial researchers working on datasets and evaluators for AI safety.
Concerns regarding the safety of AI emerge from the potential harmful uses or consequences of AI outputs, which can result from inaccuracies, irresponsibility, or inappropriate applications. As AI becomes increasingly integrated into critical systems and everyday activities, addressing AI safety issues is imperative. The misuse of AI technologies for generating misinformation, conducting sophisticated cyberattacks, developing weapons, or providing harmful advice presents grave concerns.
AI can cause societal harm, encouraging radicalisation and promoting biased or skewed views. AI-generated fake, yet highly realistic content could reduce public trust in information and government bodies. Moreover, long-term existential risks associated with the development of superintelligent AI systems cannot be ignored.
A significant portion of these safety concerns can be attributed to data-related problems at various stages of the AI lifecycle. The growing adoption of frontier foundation models in mainstream applications has amplified these concerns. Specifically, the lack of transparency regarding the data used to pre-train these models and the data approaches to fine-tuning these models for custom applications can lead to unintended consequences.
Characteristics of AI systems that need to be evaluated to ensure their safety include, but are not limited to, alignment, robustness to adversarial attacks, fairness, trustworthiness, deception capabilities, AI drift, explainability, privacy preservation, and reliability. Evaluation of these characteristics is challenging, not less due to the lack of benchmarks that are able to certify the level of safety of a given AI system.
This workshop will explore the role of data in AI safety, with a particular emphasis on data-centric AI approaches and their current limitations across the AI lifecycle.
Topics
This workshop welcomes contributions towards creation and improvement of datasets and benchmarks used for evaluation of safety of AI systems (in particular LLMs) according to the characteristics listed above. The work can be theoretical (new approaches to creating datasets or proving a dataset's adequacy for the task) or practical (suggesting new datasets, checking the validity of existing datasets, improving existing datasets). We also welcome contributions that use existing benchmarks to demonstrate safety or AI systems or find unsafe behaviours.
Format of Workshop
The workshop’s length is one full day.
The workshop will be a combination of invited talks, contributed talks, a panel, the winners and the runner-ups of the Kaggle challenge on AI Safety, presentations of submitted work, and poster presentations.
We will leave ample time for questions and answers at the end of each invited talk and panel session, and we will feature a poster session to encourage dialogue among authors.
The Kaggle challenge overview will be given by D. Sculley, the CEO of Kaggle and one of the workshop's co-chairs.
Attendance
The presenters of accepted papers and posters are expected to attend in person, except in special circumstances.
The invited speakers will attend in person.
The workshop welcomes participants that do not present a paper or a poster as well.
We expect around 100 participants in person. We will make arrangements for remote attendees.
Organizing Committee
Hana Chockler, King’s College London and the Open Data Institute (Workshop Co-Chair)
Frederik Mallmann-Trenn, King’s College London and the Open Data Institute (Workshop Co-Chair)
D. Sculley, Kaggle
Lilith Bat-Leah, Mod Op
Program Committee
Saurav Sahay (Intel Labs)
Rajat Ghosh (Nutanix)
Anka Reuel (Stanford University)
Carlos Mougan (Alan Turing Institute)
Joseph Marvin Imperial (University of Bath)
Ying-Jung Chen (Georgia Tech)
Wenhui Zhang (Bytedance)
Sean McGregor (Digital Safety Research)
Hashim Shaik (National University)
Armstrong Foundjem (Polytechnique Montreal)
Fedor Zhdanov (Nebius AI)
Nikita Agarwal (Mayo Clinic)
Sujata Goswami (Oak Ridge National Laboratory)
Usman Gohar (Iowa State University)
Satyapriya Krishna (Harvard University)
Foutse Khomh (Polytechnique Montreal)
Felix Juefei-Xu (Meta AI)
Priyanka Mary Mammen (University of Massachusetts Amherst)
Satyapriya Krishna (Harvard University)
Karan Saxena (Meta AI)
Canyu Chen (Illinois Institute of Technology)
Siddhartha Reddy Jonnalagadda (Google)
Ankit Jain (Meta Genai Safety)
Vamsi Sistla (AI Startup)
Esben Kran (Apart Research)
Ananya Gangavarapu (Ethriva)
Vamsi Sistla (MLC AI Safety WC/CTO)
Sang Truong (Stanford University)
Rebecca Weiss (MLCommons)
Dr. Raghuram Srinivas (MetricStream/SMU)
Akanksha Atrey (Nokia Bell Labs)
Xianjun Yang (University of California, Santa Barbara)
Natan Vidra (Anote)
Felix Friedrich (TU Darmstadt, Hessian.AI)
Heather Frase (MLCommons)
Eda Okur (Intel Labs)
Lily Zhang (Ford Motor Company, Latitude AI)
Zekun Wu (Holistic AI)
Patrick Schramowski (DFKI, HessianAI)
Natan Vidra (Anote)
Alice Schoenauer Sebag (Cohere)
Harit Vishwakarma (University of Wisconsin-Madison)
Amin Nikanjam (Polytechnique Montréal)
Umang Pandey (Clinica Universidad de Navarra)
Paul Röttger (Bocconi University)
Biswadeep Chakraborty (Georgia Institute of Technology)
Victor Akinwande (Carnegie Mellon)
Sabrina Hsueh (Pfizer Inc)
Virendra Mehta (University of Trento)
Trupti Bavalatti (Meta)
Masoud Charkhabi (Zoox)
Registration
Please register via AAAI: https://aaai.org/conference/aaai/aaai-25/
Call for papers
Please find the call for papers here: Call for Papers