Driven by its disciplinary diversity, AAAI has incubated numerous AI sub-disciplines and conferences and has nurtured for decades the cohesion of AI. New communities often emerge when two or more disciplines come together in order to explore new opportunities and perspectives; today both are plentiful. The purpose of this year’s Bridge Program is to tap into new sources of innovation by cultivating sustained collaboration between two or more communities, directed towards a common goal. Our interpretation of bridge is broad and encompasses disciplines both within and outside of AI. Hence, the communities that our Bridge Program is intended to bring together could be distinct subfields of AI, such as planning and learning, or different disciplines that contribute to and benefit from AI, such as AI and the humanities.
TopicsAAAI-23 welcomes submissions reporting research that advances artificial intelligence, broadly conceived. The conference scope includes machine learning (deep learning, statistical learning, etc), natural language processing, computer vision, data mining, multiagent systems, knowledge representation, human-in-the-loop AI, search, planning, reasoning, robotics and perception, and ethics. In addition to fundamental work focused on any one of these areas we expressly encourage work that cuts across technical areas of AI, (e.g., machine learning and computer vision; computer vision and natural language processing; or machine learning and planning), bridges between AI and a related research area (e.g., neuroscience; cognitive science) or develops AI techniques in the context of important application domains, such as healthcare, sustainability, transportation, and commerce.
The set of AAAI-23 keywords is available on the AAAI-23 keywords page. This author’s guide for choosing the best keywords describes important considerations in selecting keywords for a paper.
Most papers in AAAI-23 will be part of the main track. All main track papers will be reviewed according to the same criteria and via the same process. This conference has two additional tracks, which focus on AI for Social Impact and Safe and Robust AI. Papers in these two tracks will be reviewed according to a different evaluation rubric than papers in the main track. The same reviewing schedule will be followed for all papers.
Track on AI for Social Impact
As in past years, AAAI-23 will include a Track on AI for Social Impact (AISI). Submissions to this track will be reviewed according to a rubric that emphasizes the fit between the techniques used and a problem of social importance, rather than simply rewarding technical novelty. In particular, reviewers will assess significance of the problem being addressed; the paper’s engagement with previous literature on the application problem (whether in the AI literature or elsewhere); both novelty of and justification for the proposed AI-based approach; quality of evaluation; facilitation of follow-up work; and overall scope and promise for social impact. Further details are available at the AISI page.
Authors should think carefully about which track is most appropriate for their work. For example, a paper that emphasizes methodological contributions and makes weak connections to a social impact application may stand more chance of acceptance in the main technical track.
Track on Safe and Robust AI
This year we are introducing a new special track on AI systems that are safe and robust. AI is increasingly being deployed throughout society. To ensure that this technology is trustworthy, it needs to be robust to disturbance, failure and novel circumstances. Furthermore, the technology needs to offer assurance that it will reasonably avoid unsafe, irrecoverable situations. Submissions to this track will be reviewed according to a rubric that emphasizes the fit between the driving requirements of safety and robustness for AI systems and the methods and formalisms presented. Further details are available at the SRAI page.
Authors should think carefully about which track is most appropriate for their work. For example, a paper that emphasizes methodological contributions and makes weak connections to a social impact application or weak connections to problems fundamental to safety and robustness may stand more chance of acceptance in the main technical track.
Review Criteria
AAAI is a highly selective conference. Prospective authors are therefore strongly encouraged to submit only their very best work to AAAI.
All submissions will be rigorously evaluated by expert reviewers to assess whether the contributions of the paper are substantive enough to warrant publication in AAAI.
The contributions may be theoretical, methodological, algorithmic, empirical, or integrative (connecting ideas and methods across disparate subfields of AI), or critical (e.g., principled analyses and arguments that draw attention to fundamentally wrong choice of goals, assumptions, or approaches). Evaluation of submissions to the AI for Social Impact track should emphasize demonstrated or potential impact of the research in addressing pressing societal challenges, e.g., health, food, environment, education, governance, among others. Evaluation of submissions to the Safe and Robust AI track should demonstrate focussed, tangible progress towards AI systems that are safe and robust, either at the theoretical and practical level.
All submissions will be evaluated and scored for the significance and novelty of the contributions (research problems or questions addressed, methods, experiments, analyses), theoretical and/or empirical soundness of the claims, their relevance to the AAAI community, and clarity of exposition.
Additional considerations include adherence to responsible research practices (e.g., with respect to human subject studies, use of sensitive data, as well as data and algorithmic bias), and of steps to ensure reproducibility of research results (e.g., by providing detailed proofs, documenting experiments, sharing data, and sharing code).
Solid, technical papers that explore new territory or point out new directions for research, or introduce new problems, address research questions, or introduce methods that are of interest beyond a single sub-area of AI are preferred to papers that advance the state of the art, but only incrementally, or only within a narrow sub-area of AI.
Detailed instructions
More specific details are on the following pages (available soon):
Submission instructionsKeywordsSupplementary materialReview processAI for Social Impact TrackNeurIPS Fast TrackSafe and Robust AI TrackPaper modification guidelinesPaper publication and conference attendanceAffiliated eventsQuestions and SuggestionsPlease send queries about conference registration to aaai23@aaai.org. Please direct inquiries about the submission process and the technical program to workflowchairs23@aaai.zendesk.com and pcchairs23@aaai.zendesk.com respectively. Other general inquiries to the AAAI-23 General and Program Cochairs can be directed to aaai23chairs@aaai.org.
AAAI-23 Conference OrganizersGeneral Chair
Brian Williams (Massachusetts Institute of Technology, USA)
Associate General Chair
Sara Bernardini (Royal Holloway, University of London, UK)
Program Co-chairs
Yiling Chen (Harvard University, USA)Jennifer Neville (Microsoft Research and Purdue University, USA)
Associate Program Co-chairs
Michael Bowling (University of Alberta, Canada)Kai-Wei Chang (University of California, Los Angeles, USA)Daniel Lowd (University of Oregon, USA)Balaraman Ravindran (Indian Institute of Technology, Madras, India)
Workflow Co-chairs
Sungkweon Hong, (Massachusetts Institute of Technology, USA)Liunian Harold Li, (University of California Los Angeles, USA)Ying-Chun Lin, (Purdue University, USA)John Martin, (University of Alberta, Canada)Jieyu Zhao, (University of Maryland, College Park, USA)
Reproducibility Chair
Edward Raff, (Booz Allen Hamilton, USA)
IAAI-23 Conference Cochairs
Karen Haigh (Consultant, USA)Alex Wong (University of Waterloo, Canada)YuHao Chen (University of Waterloo, Canada)
EAAI-23 Symposium Cochairs
Marion Neumann (Washington University in St. Louis, USA)Pat Virtue (Carnegie Mellon University, USA)Michael Guerzhoy (University of Toronto, Canada)