The Workshop Program of the Association for the Advancement of Artificial Intelligence's 37th Conference on Artificial Intelligence (AAAI-23) was held in Washington, DC, USA on February 13-14, 2023. There were 32 workshops in the program: AI for Agriculture and Food Systems, AI for Behavior Change, AI for Credible Elections: A Call to Action with Trusted AI, AI for Energy Innovation, AI for Web Advertising, AI to Accelerate Science and Engineering, AI4EDU: AI for Education, Artificial Intelligence and Diplomacy, Artificial Intelligence for Cyber Security (AICS), Artificial Intelligence for Social Good (AI4SG), Artificial Intelligence Safety (SafeAI), Creative AI Across Modalities, Deep Learning on Graphs: Methods and Applications (DLG-AAAI'23), DEFACTIFY: Multimodal Fact-Checking and Hate Speech Detection, Deployable AI (DAI), DL-Hardware Co-Design for AI Acceleration, Energy Efficient Training and Inference of Transformer Based Models, Graphs and More Complex Structures for Learning and Reasoning (GCLR), Health Intelligence (W3PHIAI-23), Knowledge-Augmented Methods for Natural Language Processing, Modelling Uncertainty in the Financial World (MUFin'23), Multi-Agent Path Finding, Multimodal AI for Financial Forecasting (Muffin), Multimodal AI for Financial Forecasting (Muffin), Privacy-Preserving Artificial Intelligence, Recent Trends in Human-Centric AI, Reinforcement Learning Ready for Production, Scientific Document Understanding, Systems Neuroscience Approach to General Intelligence, Uncertainty Reasoning and Quantification in Decision Making (UDM'23), User-Centric Artificial Intelligence for Assistance in At-Home Tasks, and When Machine Learning Meets Dynamical Systems: Theory and Applications. This report contains summaries of the workshops, which were submitted by some, but not all of the workshop chairs. An increasing world population, coupled with finite arable land, changing diets, and the growing expense of agricultural inputs, is poised to stretch our agricultural systems to their limits. By the end of this century, the earth's population is projected to increase by 45% with available arable land decreasing by 20% coupled with changes in what crops these arable lands can best support; this creates the urgent need to enhance agricultural productivity by 70% before 2050.

The symposium consisted of four keynote speakers with varying engagement with the idea of a distributed teaching collaborative. The symposium began with a keynote by Chad Jenkins, a professor of robotics at the University of Michigan. In his talk, he laid out the symposium's goal and the intent of the distributed teaching collaborative. He highlighted efforts to offer distributed classes between the University of Michigan and Barea College, Howard University, and Morehouse College. The second keynote was Dr. Talitha Washington the current director of the AUC data science initiative.

The audience was then split into four groups, with each faculty presenter leading the group for interactive group activities. Following student introductions about their educational background and reason for attending the workshop, each group was given a very short peer-reviewed research paper on topics ranging from natural language processing, search, human-automation interaction, and multiagent systems to read with guidance from the faculty presenter on how to read a research paper, students read each section of the paper and discussed it with their group in response to questions from the workbook. The faculty then shared a few specific research projects from their own research areas and introduced undergraduate summer research programs. The last part of the group activity time involved a discussion on how students could seek out and secure research opportunities. This included specifics on how to prepare and reach out to faculty members about opportunities to do research in their labs.

This year was the first in-person EXAG since the start of the COVID-19 pandemic. We did our best to support a hybrid event to accommodate international presenters. We had excellent attendance on both days, with five paper sessions and two demo sessions (one formal, one informal). Our presentations spanned the following themes: story world generation, level generation, pixel art generation, adaptive MCTS, open-endedness in games, level reachability testing, NPC behaviors, AI-driven sonification, unit generation for real-time strategy games, empathetic AI, expressive range visualization, emulator frameworks for MCTS, and reinforcement Learning for fighting game AI. This year, EXAG received 22 submissions (21 for the research track and 1 for the demo track).

For ASR, this milestone was first claimed in a 2016 research paper by Microsoft (Xiong et al., 2016) reporting that for the first time, they have achieved human parity in word error rate1 (WER) on the Switchboard benchmark (5.8% WER) while also achieving 11% WER on the CallHome benchmark, which is known to be more challenging to transcribe. In addition, the reported decoding speed was only 1.38 real time, which is in the realm of usability for some commercial systems. This announcement was highly publicized even in mainstream media outlets2. A follow-up paper in 2017 claimed further improvement to 5.1% WER on Switchboard but with no report on decoding speed (Xiong et al., 2018). Also in 2017, Google announced a 4.9% WER (on some undisclosed benchmark) at its annual I/O developer conference3.

Do It Yourself, but Not Alone: Companion-Technology for Home Improvement – Bringing a Planning-Based Interactive DIY Assistant to Life.

Do It Yourself, but Not Alone: Com- panion-Technology for Home Improvement – Bringing a Planning-Based Interactive DIY Assistant to Life.

The following article is an extended abstract submitted as part of AAAI's New Faculty Highlights Program. With the advances of deep learning techniques, text generation is attracting increasing interest in the artificial intelligence (AI) commu- nity, because of its wide applications and because it is an essential component of AI. Traditional text generation systems are trained in a supervised way, requiring massive labeled parallel corpora. In this paper, I will introduce our recent work on search and learning ap- proaches to unsupervised text generation, where a heuristic objective function estimates the quality of a candidate sentence, and discrete search algorithms generate a sentence by maximizing the search objective. A machine learning model further learns from the search results to smooth out noise and improve efficiency.

The following article is an extended abstract submitted as part of AAAI's New Faculty Highlights Program. Recent years have witnessed a surge in machine learning methods on graph data, especially those pow- ered by effective neural networks. Despite their success in different real-world scenarios, the majority of these methods on graphs only focus on predictive or descriptive tasks, while they lack any perspective of causality. Causal inference can reveal the causality inside data. An important problem in causal inference is causal effect estimation, which aims to estimate the causal effects of a certain treatment (e.g., prescrip- tion of medicine) on an outcome (e.g., cure of disease).