Table of Contents

• Position: Rethinking LLM Bias Probing Using Lessons from the Social SciencesKirsten N. Morehouse, Siddharth Swaroop, Weiwei Pan. [doi]
• Position: Machine Learning Models Have a Supply Chain ProblemSarah Meiklejohn, Hayden Blauzvern, Mihai Maruseac, Spencer Schrock, Laurent Simon, Ilia Shumailov. [doi]
• Position: A Theory of Deep Learning Must Include Compositional SparsityDavid A. Danhofer, Davide D'Ascenzo, Rafael Dubach, Tomaso A. Poggio. [doi]
• Position: It Is Time We Test Neural Computation In VitroFrithjof Gressmann, Ashley Chen, Lily Hexuan Xie, Nancy M. Amato, Lawrence Rauchwerger. [doi]
• Position: Stop treating 'AGI' as the north-star goal of AI researchBorhane Blili-Hamelin, Christopher Graziul, Leif Hancox-Li, Hananel Hazan, El Mahdi El Mhamdi, Avijit Ghosh, Katherine A. Heller, Jacob Metcalf, Fabricio Murai, Eryk Salvaggio, Andrew Smart, Todd Snider, Mariame Tighanimine, Talia Ringer, Margaret Mitchell, Shiri Dori-Hacohen. [doi]
• Position: Not All Explanations for Deep Learning Phenomena Are Equally ValuableAlan Jeffares, Mihaela van der Schaar. [doi]
• Position: Truly Self-Improving Agents Require Intrinsic Metacognitive LearningTennison Liu, Mihaela van der Schaar. [doi]
• Position: AI Safety should prioritize the Future of WorkSanchaita Hazra, Bodhisattwa Prasad Majumder, Tuhin Chakrabarty. [doi]
• Position: Generative AI Regulation Can Learn from Social Media RegulationRuth Elisabeth Appel. [doi]
• Position: Humanity Faces Existential Risk from Gradual DisempowermentJan Kulveit, Raymond Douglas, Nora Ammann, Deger Turan, David Krueger 0001, David Duvenaud. [doi]
• Position: Beyond Assistance - Reimagining LLMs as Ethical and Adaptive Co-Creators in Mental Health CareAbeer Badawi, Md. Tahmid Rahman Laskar, Jimmy Huang 0001, Shaina Raza, Elham Dolatabadi. [doi]
• Position: Evaluating Generative AI Systems Is a Social Science Measurement ChallengeHanna M. Wallach, Meera A. Desai, A. Feder Cooper, Angelina Wang, Chad Atalla, Solon Barocas, Su Lin Blodgett, Alexandra Chouldechova, Emily Corvi, P. Alex Dow, Jean Garcia-Gathright, Alexandra Olteanu, Nicholas J. Pangakis, Stefanie Reed, Emily Sheng, Dan Vann, Jennifer Wortman Vaughan, Matthew Vogel, Hannah Washington, Abigail Z. Jacobs. [doi]
• Position: Iterative Online-Offline Joint Optimization is Needed to Manage Complex LLM Copyright RisksYanzhou Pan, Jiayi Chen, Jiamin Chen, Zhaozhuo Xu, Denghui Zhang. [doi]
• Position: Language model developers should report train-test overlapAndy K. Zhang, Kevin Klyman, Yifan Mai 0001, Yoav Levine, Yian Zhang, Rishi Bommasani, Percy Liang. [doi]
• Position: AI's growing due process problemSunayana Rane. [doi]
• Position: AI Competitions Provide the Gold Standard for Empirical Rigor in GenAI EvaluationD. Sculley, William Cukierski, Phil Culliton, Sohier Dane, Maggie Demkin, Ryan Holbrook, Addison Howard, Paul Mooney, Walter Reade, Meg Risdal, Nate Keating. [doi]
• Position: AI Safety Must Embrace an Antifragile PerspectiveMing Jin 0002, Hyunin Lee. [doi]
• Position: AI Agents Need Authenticated DelegationTobin South, Samuele Marro, Thomas Hardjono, Robert Mahari, Cedric Deslandes Whitney, Alan Chan, Alex Pentland. [doi]
• Position: Explainable AI Cannot Advance Without Better User StudiesMatej Piculin, Bernarda Petek, Irena Ograjensek, Erik Strumbelj. [doi]
• Position: Challenges and Future Directions of Data-Centric AI AlignmentMin-Hsuan Yeh, Jeffrey Wang, Xuefeng Du, Seongheon Park, Leitian Tao, Shawn Im, Yixuan Li 0001. [doi]
• Position: Human Baselines in Model Evaluations Need Rigor and Transparency (With Recommendations & Reporting Checklist)Kevin L. Wei, Patricia Paskov, Sunishchal Dev, Michael J. Byun, Anka Reuel, Xavier Roberts-Gaal, Rachel Calcott, Evie Coxon, Chinmay Deshpande. [doi]
• Position: Causal Machine Learning Requires Rigorous Synthetic Experiments for Broader AdoptionAudrey Poinsot, Panayiotis Panayiotou 0002, Alessandro Ferreira Leite, Nicolas Chesneau, Özgür Simsek, Marc Schoenauer. [doi]
• Position: Current Model Licensing Practices are Dragging Us into a Quagmire of Legal NoncomplianceMoming Duan, Mingzhe Du, Rui Zhao 0009, Mengying Wang 0001, Yinghui Wu 0001, Nigel Shadbolt, Bingsheng He. [doi]
• Position: AI Scaling: From Up to Down and OutYunke Wang, Yanxi Li 0001, Chang Xu 0002. [doi]
• Position: AI Evaluation Should Learn from How We Test HumansYan Zhuang, Qi Liu 0003, Zachary A. Pardos, Patrick C. Kyllonen, Jiyun Zu, Zhenya Huang, Shijin Wang 0001, Enhong Chen. [doi]
• Position: Algebra Unveils Deep Learning - An Invitation to Neuroalgebraic GeometryGiovanni Luca Marchetti, Vahid Shahverdi, Stefano Mereta, Matthew Trager, Kathlén Kohn. [doi]
• Position: The Right to AIRashid Mushkani, Hugo Berard, Allison Cohen, Shin Koseki. [doi]
• Position: The Artificial Intelligence and Machine Learning Community Should Adopt a More Transparent and Regulated Peer Review ProcessJing Yang. [doi]
• Position: Lifetime tuning is incompatible with continual reinforcement learningGolnaz Mesbahi, Parham Mohammad Panahi, Olya Mastikhina, Steven Tang, Martha White, Adam White 0001. [doi]
• Position: Rethinking Explainable Machine Learning as Applied StatisticsSebastian Bordt, Eric Raidl, Ulrike von Luxburg. [doi]
• Position: Enough of Scaling LLMs! Lets Focus on DownscalingYash Goel, Ayan Sengupta, Tanmoy Chakraborty 0002. [doi]
• Position: We Need Responsible, Application-Driven (RAD) AI ResearchSarah Hartman, Cheng Soon Ong, Julia Powles, Petra Kuhnert. [doi]
• Position: Democratic AI is Possible. The Democracy Levels Framework Shows How It Might WorkAviv Ovadya, Kyle Redman, Luke Thorburn, Quan Ze Chen, Oliver Smith, Flynn Devine, Andrew Konya, Smitha Milli, Manon Revel, Kevin Feng, Amy X. Zhang, Bilva Chandra, Michiel A. Bakker, Atoosa Kasirzadeh. [doi]
• Position: AI Should Not Be An Imitation Game: Centaur EvaluationsAndreas Haupt, Erik Brynjolfsson. [doi]
• Position: Supervised Classifiers Answer the Wrong Questions for OOD DetectionYucen Lily Li, Daohan Lu, Polina Kirichenko, Shikai Qiu, Tim G. J. Rudner, C. Bayan Bruss, Andrew Gordon Wilson. [doi]
• Position: General Intelligence Requires Reward-based PretrainingSeungwook Han, Jyothish Pari, Samuel J. Gershman, Pulkit Agrawal 0001. [doi]
• Position: Strong Consumer Protection is an Inalienable Defense for AI Safety in the United StatesSerena Booth. [doi]
• Position: We Can't Understand AI Using our Existing VocabularyJohn Hewitt, Robert Geirhos, Been Kim. [doi]
• Position: Deep Learning is Not So Mysterious or DifferentAndrew Gordon Wilson. [doi]
• Position: Retrieval-augmented systems can be dangerous medical communicatorsLionel Wong, Ayman Ali, Raymond M. Xiong, Zejiang Shen 0001, Yoon Kim, Monica Agrawal. [doi]
• Position: Societal Impacts Research Requires Benchmarks for Creative Composition TasksJudy Hanwen Shen. [doi]
• Position: Contextual Integrity is Inadequately Applied to Language ModelsYan Shvartzshnaider, Vasisht Duddu. [doi]
• Position: Medical Large Language Model Benchmarks Should Prioritize Construct ValidityAhmed Alaa, Thomas Hartvigsen, Niloufar Golchini, Shiladitya Dutta, Frances Dean, Inioluwa Deborah Raji, Travis Zack. [doi]
• Position: Probabilistic Modelling is Sufficient for Causal InferenceBruno Kacper Mlodozeniec, David Krueger 0001, Richard E. Turner. [doi]
• Position: Future Research and Challenges Remain Towards AI for Software EngineeringAlex Gu, Naman Jain, Wen-Ding Li, Manish Shetty, Kevin Ellis, Koushik Sen, Armando Solar-Lezama. [doi]
• Position: An Empirically Grounded Identifiability Theory Will Accelerate Self Supervised Learning ResearchPatrik Reizinger, Randall Balestriero, David A. Klindt, Wieland Brendel. [doi]
• Position: Certified Robustness Does Not (Yet) Imply Model SecurityAndrew Craig Cullen, Paul Montague, Sarah Monazam Erfani, Benjamin I. P. Rubinstein. [doi]
• Position: The Future of Bayesian Prediction Is Prior-FittedSamuel Müller 0005, Arik Reuter, Noah Hollmann, David Rügamer, Frank Hutter. [doi]
• Position: Graph Matching Systems Deserve Better BenchmarksIndradyumna Roy, Saswat Meher, Eeshaan Jain, Soumen Chakrabarti, Abir De. [doi]
• Position: The Categorization of Race in ML is a Flawed PremiseMiriam Doh, Benedikt Höltgen, Piera Riccio, Nuria Oliver. [doi]
• Position: Build Agent Advocates, Not Platform AgentsSayash Kapoor, Noam Kolt, Seth Lazar. [doi]
• Position: In-House Evaluation Is Not Enough. Towards Robust Third-Party Evaluation and Flaw Disclosure for General-Purpose AIShayne Longpre, Kevin Klyman, Ruth Elisabeth Appel, Sayash Kapoor, Rishi Bommasani, Michelle Sahar, Sean McGregor, Avijit Ghosh, Borhane Blili-Hamelin, Nathan Butters, Alondra Nelson, Amit Elazari, Andrew Sellars, Casey John Ellis, Dane Sherrets, Dawn Song, Harley Geiger, Ilona Cohen, Lauren McIlvenny, Madhulika Srikumar, Mark M. Jaycox, Markus Anderljung, Nadine Farid Johnson, Nicholas Carlini, Nicolas Miailhe, Nik Marda, Peter Henderson 0002, Rebecca S. Portnoff, Rebecca Weiss, Victoria Westerhoff, Yacine Jernite, Rumman Chowdhury, Percy Liang, Arvind Narayanan. [doi]
• Position: The Most Expensive Part of an LLM *should* be its Training DataNikhil Kandpal, Colin Raffel. [doi]
• Position: Trustworthy AI Agents Require the Integration of Large Language Models and Formal MethodsYedi Zhang, Yufan Cai 0001, Xinyue Zuo, Xiaokun Luan, Kailong Wang 0001, Zhe Hou, Yifan Zhang, Zhiyuan Wei, Meng Sun 0002, Jun Sun 0001, Jing Sun 0002, Jin Song Dong 0001. [doi]
• Position: Scaling LLM Agents Requires Asymptotic Analysis with LLM PrimitivesElliot Meyerson, Xin Qiu 0001. [doi]
• Position: Uncertainty Quantification Needs Reassessment for Large Language Model AgentsMichael Kirchhof 0002, Gjergji Kasneci, Enkelejda Kasneci. [doi]
• Position: When Incentives Backfire, Data Stops Being HumanSebastin Santy, Prasanta Bhattacharya, Manoel Horta Ribeiro, Kelsey R. Allen, Sewoong Oh. [doi]
• Position: Solve Layerwise Linear Models First to Understand Neural Dynamical Phenomena (Neural Collapse, Emergence, Lazy/Rich Regime, and Grokking)Yoonsoo Nam, Seok Hyeong Lee, Clémentine Carla Juliette Dominé, Yeachan Park, Charles London, Wonyl Choi, Niclas Alexander Göring, Seungjai Lee. [doi]
• Position: Theory of Mind Benchmarks are Broken for Large Language ModelsMatthew Riemer, Zahra Ashktorab, Djallel Bouneffouf 0001, Payel Das, Miao Liu 0001, Justin D. Weisz, Murray Campbell. [doi]
• Position: Constants are Critical in Regret Bounds for Reinforcement LearningSimone Drago, Marco Mussi, Alberto Maria Metelli. [doi]
• Position: Spectral GNNs Rely Less on Graph Fourier Basis than ConceivedYuhe Guo, Huayi Tang, JiaHong Ma, Hongteng Xu, Zhewei Wei. [doi]
• Position: Graph Learning Will Lose Relevance Due To Poor BenchmarksMaya Bechler-Speicher, Ben Finkelshtein, Fabrizio Frasca, Luis Müller, Jan Tönshoff, Antoine Siraudin, Viktor Zaverkin, Michael M. Bronstein, Mathias Niepert, Bryan Perozzi, Mikhail Galkin 0001, Christopher Morris 0001. [doi]
• Position: Principles of Animal Cognition to Improve LLM EvaluationsSunayana Rane, Cyrus F. Kirkman, Graham Todd, Amanda L. Royka, Ryan M. C. Law, Erica A. Cartmill, Jacob Gates Foster. [doi]
• Position: Formal Mathematical Reasoning - A New Frontier in AIKaiyu Yang, Gabriel Poesia, Jingxuan He, Wenda Li, Kristin E. Lauter, Swarat Chaudhuri, Dawn Song. [doi]
• Position: Editing Large Language Models Poses Serious Safety RisksPaul Youssef, Zhixue Zhao, Daniel Braun 0013, Jörg Schlötterer, Christin Seifert. [doi]
• Position: Political Neutrality in AI Is Impossible - But Here Is How to Approximate ItJillian Fisher, Ruth Elisabeth Appel, Chan Young Park, Yujin Potter, Liwei Jiang, Taylor Sorensen, Shangbin Feng, Yulia Tsvetkov, Margaret E. Roberts, Jennifer Pan, Dawn Song, Yejin Choi 0001. [doi]
• Position: LLM Social Simulations Are a Promising Research MethodJacy Reese Anthis, Ryan Liu 0001, Sean M. Richardson, Austin C. Kozlowski, Bernard Koch, Erik Brynjolfsson, James A. Evans, Michael S. Bernstein. [doi]
• Position: You Can't Manufacture a NeRFMa Kimmel, Mueed Ur Rehman, Yonatan Bisk, Gary K. Fedder. [doi]
• Position: All Current Generative Fidelity and Diversity Metrics are FlawedOssi Räisä, Boris van Breugel, Mihaela van der Schaar. [doi]
• Position: The AI Conference Peer Review Crisis Demands Author Feedback and Reviewer RewardsJaeho Kim, Yunseok Lee, Seulki Lee. [doi]
• Position: Don't Use the CLT in LLM Evals With Fewer Than a Few Hundred DatapointsSam Bowyer, Laurence Aitchison, Desi R. Ivanova. [doi]
• Position: LLMs Need a Bayesian Meta-Reasoning Framework for More Robust and Generalizable ReasoningHanqi Yan, Linhai Zhang, Jiazheng Li 0002, Zhenyi Shen, Yulan He 0001. [doi]
• Position: We Need An Algorithmic Understanding of Generative AIOliver Eberle, Thomas McGee, Hamza Giaffar, Taylor Whittington Webb, Ida Momennejad. [doi]