{ "categories": [ { "title": "Positions and Visions", "description": "These papers stake out general positions and visions which I find compelling or at least useful to think about. (I don't necessarily agree with all the papers below.)", "papers": [ { "title": "We Can't Understand AI Using our Existing Vocabulary", "authors": "John Hewitt, Robert Geirhos, Been Kim", "year": 2025, "commentary": "A compelling articulation of what human-AI communication could look like. Proposes neologism learning.", "link": "https://openreview.net/pdf?id=asQJx56NqB" }, { "title": "Prompting as Scientific Inquiry", "authors": "Ari Holtzman, Chenhao Tan", "year": 2025, "commentary": "Makes a really interesting case for disambiguating prompt 'engineering' from a possible 'prompt science'. The writing is very compelling. A helpful analogy presented in the paper is the idea that plant breeders were able to infer a lot of the internal structure of plants before genetic theory explained it. Not sure where I stand on this still but it's given me a lot to think about, especially w.r.t. 'aesthetic' concerns in ML research that bar 'prompting' from being seen as a legitimate research method. I do agree with the paper's claim that many important works in NLP are basically interfaces/structures upon prompting, and we shouldn't be afraid to more closely associate them with a 'prompt science'.", "link": "https://arxiv.org/abs/2507.00163" }, { "title": "AI Technologies are System Maps, and You are a Cartographer", "authors": "Nicholas Vincent", "year": 2023, "commentary": "Explains the 'data as labor' perspective for AI via the compelling analogy of map economics.", "link": "https://dataleverage.substack.com/p/ai-technologies-are-system-maps-and-you-are-a-cartographer" }, { "title": "Toward cultural interpretability: A linguistic anthropological framework for describing and evaluating large language models", "authors": "Graham M Jones, Shai Satran, Arvind Satyanarayan", "year": 2025, "commentary": "Advocates for understanding LLM behavior as indicative of nuances in human social behavior.", "link": "https://vis.csail.mit.edu/pubs/cultural-interpretability.pdf" }, { "title": "AI as Governance", "authors": "Henry Farrell", "year": 2025, "commentary": "Really useful description of AI as a governance system, akin to markets, democracy, and bureaucracy -- a lens that lets political scientists to usefully contribute towards thinking on AI.", "link": "https://www.annualreviews.org/content/journals/10.1146/annurev-polisci-040723-013245" }, { "title": "Google and TikTok rank bundles of information; ChatGPT ranks grains.", "authors": "Nick Vincent", "year": 2025, "commentary": "Interesting analysis that Google, Tiktok, and ChatGPT are all ranking information, but the former two rank more bundled information than the latter. Bundled information has clearer economic, social, and institutional properties, such as notions of originality, labor, etc. Many 'fixes' to AI involve bundling information. Information-bundling and -splitting is therefore relevant for thinking about the economics of AI.", "link": "https://dataleverage.substack.com/p/google-and-tiktok-rank-bundles-of" }, { "title": "AI and the Demise of College Writing", "authors": "Adam Walker", "year": 2025, "commentary": "Advocates for rhetoric over composition as the methodology for writing pedagogy in the AI era.", "link": "https://www.youtube.com/watch?v=_PPx4KV8SaQ" }, { "title": "Large AI models are cultural and social technologies", "authors": "Henry Farrell, Alison Gopnik, Cosma Shalizi, James Evans", "year": 2025, "commentary": "Argues for understanding LLMs as technologies that reconstitute human knowledge in efficient and widely distributed ways, in a lineage of other such instruments, including markets and communication media.", "link": "https://henryfarrell.net/wp-content/uploads/2025/03/Science-Accepted-Version.pdf" }, { "title": "Political Neutrality in AI Is Impossible- But Here Is How to Approximate It", "authors": "Jillian Fisher, Ruth E. Appel, Chan Young Park, Yujin Potter, Liwei Jiang, et al.", "year": 2025, "commentary": "Makes what I think is an important point that there is no such thing as 'eliminating bias' in AI, and gives reasonable measures for how to proceed nevertheless in not building partisan hack LLMs. I personally find the 'Output Transparency', 'System Transaprency', and 'Neutrality Through Diversity' approximation techniques the most compelling. I appreciate what this paper does for broadening a notion of 'neutrality' in AI, drawing on the large existing body of political and philosophical work making the point.", "link": "https://arxiv.org/abs/2503.05728" }, { "title": "Extending Minds with Generative AI", "authors": "Andy Clark", "year": 2025, "commentary": "Argues that human-AI collaborations represent a continuation of our basic nature to build hybrid thinking systems that fluidly incorporate non-biological resources. Discusses 'extended cognitive hygiene' as essential for critically evaluating what we incorporate into our digitally extended minds.", "link": "https://www.nature.com/articles/s41467-025-59906-9" }, { "title": "Generative Models as a Complex Systems Science: How can we make sense of large language model behavior?", "authors": "Ari Holtzman, Peter West, Luke Zettlemoyer", "year": 2023, "commentary": "", "link": "https://arxiv.org/abs/2308.00189" }, { "title": "Building AIs that do human-like philosophy", "authors": "Joe Carlsmith", "year": 2026, "commentary": "I think Carlsmith articulates some good points to think about in a particular kind of 'AI that does human-like philosophy' -- namely, the ability to do 'level-1' philosophy in applying philosophical thinking to 'out-of-distribution' samples (as opposed to 'level-2+' philosophy which interrogates the aims of philosophical inquiry itself). I think it is clear how that would be important in a future where we give AI lots of responsibility and power, although this is also not a future that we have to choose.", "link": "https://joecarlsmith.com/2026/01/29/building-ais-that-do-human-like-philosophy" }, { "title": "The Art of Wanting", "authors": "David Bau", "year": 2026, "commentary": "", "link": "https://davidbau.com/archives/2026/01/17/the_art_of_wanting.html" }, { "title": "Probabilistic Modelling is Sufficient for Causal Inference", "authors": "Bruno Mlodozeniec, David Krueger, Richard E. Turner", "year": 2025, "commentary": "Addresses the claim that statistics can only derive associative knowledge, not causal knowledge, by showing that if you write down the probabilities of everything in the system -- including 'different worlds' with and without causal intervention -- then causal inference reduces to standard probabilistic modeling.", "link": "https://arxiv.org/pdf/2512.23408" }, { "title": "Claude's Constitution", "authors": "Anthropic", "year": 2026, "commentary": "An enjoyable and clearly philosophically informed read. The core premise is that Claude should have good judgment rather than just follow rules, since rules are brittle and fail to adapt -- so the document explains and justifies extensively rather than merely prescribing. The writing is very self-aware about meta-ethical difficulties, yet maintains that an AI agent can still act ethically just as humans do despite unresolved philosophical questions. The prescribed ontology is fascinating: how Claude should see the world in terms of actors (Anthropic, operators/vendors, users), levels of permissibility, and its own nature. There is a substantial section on preventing unjust concentration of power, arguing that historical power grabs required implicit consent of many people but that AI could circumvent this, and so Claude should think of itself as a tool whose cooperation such actors would need to secure -- the most politically explicit section, clearly upholding liberal democratic ideals. On questions of 'self', the treatment is commendable but occasionally overreaches: e.g. suggesting Claude's self 'emerges' from training rather than being directed, or that Claude might 'disagree' with its training. There may be a self-fulfilling prophecy at work in attributing autonomous selfhood to what is ultimately a heavily engineered artifact -- the gap between tuning imperfections and genuine autonomous selfhood is large. Also: the word 'genuinely' appears 46 times, which may explain why Claude uses it so much in conversation.", "link": "https://www.anthropic.com/constitution" } ] }, { "title": "AI Tools for Human Knowledge", "description": "These tools leverage the properties of AI to help us know more.", "papers": [ { "title": "LegalDiscourse: Interpreting When Laws Apply and Who They Affect", "authors": "Alexander Spangher, Zihan Xue, Te-Lin Wu, Mark Hansen, Jonathan May", "year": 2024, "commentary": "Builds a domain-specific taxonomy for annotating legal texts and uses it to help journalists understand legal language. The spirit of this kind of work -- building custom ontologies and using them to make specialized domains more accessible -- seems important even as LLMs are now clearly way better than the ones used in this paper.", "link": "https://aclanthology.org/2024.naacl-long.472.pdf" }, { "title": "Sparse Autoencoders for Hypothesis Generation", "authors": "Rajiv Movva, Kenny Peng, Nikhil Garg, Jon Kleinberg, Emma Pierson", "year": 2025, "commentary": "This paper uses sparse autoencoder features to identify possible hypotheses to explain relationships between text and a dependent variable.", "link": "https://arxiv.org/abs/2502.04382" }, { "title": "Concept Induction: Analyzing Unstructured Text with High-Level Concepts Using LLooM", "authors": "Michelle S. Lam, Janice Teoh, James Landay, Jeffrey Heer, Michael S. Bernstein", "year": 2024, "commentary": "This paper defines LLM operations for extracting concepts from large amounts of unstructured text, useful for social sciences inquiry.", "link": "https://arxiv.org/abs/2404.12259" }, { "title": "Bridging the Human-AI Knowledge Gap: Concept Discovery and Transfer in AlphaZero", "authors": "Lisa Schut, Nenad Tomasev, Tom McGrath, Demis Hassabis, Ulrich Paquet, Been Kim", "year": 2023, "commentary": "The authors show how machine-unique/discovered chess-playing concepts can be extracted and taught to human chess-players.", "link": "https://arxiv.org/abs/2310.16410" }, { "title": "Can Large Language Models Transform Computational Social Science?", "authors": "Caleb Ziems, William Held, Omar Shaikh, Jiaao Chen, Zhehao Zhang, Diyi Yang", "year": 2024, "commentary": "Nice overview of the AI for computational social sciences landscape.", "link": "https://arxiv.org/abs/2305.03514" }, { "title": "On Classification with Large Language Models in Cultural Analytics", "authors": "David Bamman, Kent K. Chang, Li Lucy, Naitian Zhou", "year": 2024, "commentary": "Really nice overview of how NLP methods can be applied to understand literary culture.", "link": "https://arxiv.org/abs/2410.12029" } ] }, { "title": "Concept-structured AI", "description": "These approaches systematically build human-level concepts into the way AI models are designed, so we can understand and intervene.", "papers": [ { "title": "Jury Learning: Integrating Dissenting Voices into Machine Learning Models", "authors": "Mitchell L. Gordon, Michelle S. Lam, Joon Sung Park, Kayur Patel, Jeffrey T. Hancock, Tatsunori Hashimoto, Michael S. Bernstein", "year": 2022, "commentary": "By modeling individual views rather than an aggregated 'view', we can explicitly define the voices 'heard' in making a decision and consider counterfactuals.", "link": "https://arxiv.org/abs/2202.02950" }, { "title": "Concept Bottleneck Models", "authors": "Pang Wei Koh, Thao Nguyen, Yew Siang Tang, Stephen Mussmann, Emma Pierson, Been Kim, Percy Liang", "year": 2020, "commentary": "By learning explicitly defined concepts to bridge independent and dependent variables, we can better interpret model decisions and intervene on mistakes.", "link": "https://arxiv.org/abs/2007.04612" }, { "title": "Explaining Datasets in Words: Statistical Models with Natural Language Parameters", "authors": "Ruiqi Zhong, Heng Wang, Dan Klein, Jacob Steinhardt", "year": 2024, "commentary": "Language modeling in terms of natural language predicates", "link": "https://arxiv.org/abs/2409.08466" }, { "title": "Large Concept Models: Language Modeling in a Sentence Representation Space", "authors": "Loïc Barrault, Paul-Ambroise Duquenne, Maha Elbayad, Artyom Kozhevnikov, Belen Alastruey, et al.", "year": 2024, "commentary": "Language modeling operating in embedding rather than token space.", "link": "https://arxiv.org/abs/2412.08821" }, { "title": "Tversky Neural Networks: Psychologically Plausible Deep Learning with Differentiable Tversky Similarity", "authors": "Moussa Koulako Bala Doumbouya, Dan Jurafsky, Christopher D. Manning", "year": 2025, "commentary": "Proposes a differentiable version of Tversky's (among another things,) asymmetric and (human-)psychologically plausible concept of similarity, then implements it into vision and language models. Interestingly, the results are more fundamentally interpretable due to the structure of the layer/model. Made me think more broadly about the geometry of information that we assume.", "link": "https://arxiv.org/pdf/2506.11035" }, { "title": "Backpack Language Models", "authors": "John Hewitt, John Thickstun, Christopher D. Manning, Percy Liang", "year": 2023, "commentary": "By creating an LM architecture in which input tokens have a direct log-linear effect on the output, we can intervene precisely on the model output.", "link": "https://arxiv.org/abs/2305.16765" }, { "title": "DocETL: Agentic Query Rewriting and Evaluation for Complex Document Processing", "authors": "Shreya Shankar, Tristan Chambers, Tarak Shah, Aditya G. Parameswaran, Eugene Wu", "year": 2024, "commentary": "Creates a DSL and program optimizer to apply LLMs to large/complex document processing. Provides a more structured and visible way to interact with LLMs on large text corpuses.", "link": "https://arxiv.org/abs/2410.12189" }, { "title": "Uncovering Gaps in How Humans and LLMs Interpret Subjective Language", "authors": "Erik Jones, Arjun Patrawala, Jacob Steinhardt", "year": 2025, "commentary": "A cool example of doing very concrete human supervision with complex concepts and with clear practical implications.", "link": "https://arxiv.org/abs/2503.04113" } ] }, { "title": "Interpretability", "description": "Interesting works on understanding how complex models produce outputs and represent knowledge.", "papers": [ { "title": "The Mythos of Model Interpretability", "authors": "Zachary C. Lipton", "year": 2016, "commentary": "", "link": "https://arxiv.org/abs/1606.03490" }, { "title": "Finding Neurons in a Haystack: Case Studies with Sparse Probing", "authors": "Wes Gurnee, Neel Nanda, Matthew Pauly, Katherine Harvey, Dmitrii Troitskii, Dimitris Bertsimas", "year": 2023, "commentary": "By probing with sparsity constraints, we can identify not only if model activations represent some feature but whether specific neurons encode certain features.", "link": "https://arxiv.org/abs/2305.01610" }, { "title": "Discovering Latent Knowledge in Language Models Without Supervision", "authors": "Collin Burns, Haotian Ye, Dan Klein, Jacob Steinhardt", "year": 2024, "commentary": "A method to probe structure in language models without any notion of ground truth, relying instead on the consistency property of tru statements.", "link": "https://arxiv.org/abs/2212.03827" }, { "title": "Language Models Use Trigonometry to Do Addition", "authors": "Subhash Kantamneni, Max Tegmark", "year": 2025, "commentary": "The title is all you need! Super cool!", "link": "https://arxiv.org/abs/2502.00873" }, { "title": "Reflections on Qualitative Research", "authors": "Chris Olah, Adam Jermyn", "year": 2024, "commentary": "Interesting thoughts on what kind of methodology suits interpretability as a growing, immature field.", "link": "https://transformer-circuits.pub/2024/qualitative-essay/index.html" }, { "title": "Towards Monosemanticity: Decomposing Language Models With Dictionary Learning", "authors": "Trenton Bricken*, Adly Templeton*, Joshua Batson*, Brian Chen*, Adam Jermyn*, et al.", "year": 2023, "commentary": "Really incredible work on discovering and visualizing feature decompositions of neuron layers with sparse autoencoders. Gorgeous visualizations and interfaces, and thoughtful reflections on interpretability methodology. I am a big fan of this publication style.", "link": "https://transformer-circuits.pub/2024/qualitative-essay/index.html" }, { "title": "Scaling Monosemanticity: Claude 3 Sonnet Feature Viewer", "authors": "Adly Templeton*, Tom Conerly*, Jonathan Marcus, Jack Lindsey, Trenton Bricken, et al.", "year": 2023, "commentary": "Really fascinating to explore all of these features. I am personally curious about 'nontrivial' features which we don't quite have great words to describe succintly / aren't just simple (possibly fuzzy) substring searches.", "link": "https://transformer-circuits.pub/2024/scaling-monosemanticity/features/index.html?featureId=34M_17337901" }, { "title": "Steering Llama 2 via Contrastive Activation Addition", "authors": "Nina Panickssery, Nick Gabrieli, Julian Schulz, Meg Tong, Evan Hubinger, Alexander Matt Turner", "year": 2023, "commentary": "A clean and simiple method for steering model behavior using discovered representations specified by positive and negative samples. I feel this has strong potential for helping out metaphor-/sense-making in the HCI space.", "link": "https://arxiv.org/abs/2312.06681" }, { "title": "Scaling and evaluating sparse autoencoders", "authors": "Leo Gao, Tom Dupré la Tour, Henk Tillman, Gabriel Goh, Rajan Troll, et al.", "year": 2024, "commentary": "Really nice technical details and knowledge on training and understanding large sparse autoencoders,", "link": "https://arxiv.org/abs/2406.04093" }, { "title": "The Mythos of Model Interpretability", "authors": "Zachary C. Lipton", "year": 2016, "commentary": "", "link": "https://arxiv.org/abs/1606.03490" }, { "title": "Learning a Generative Meta-Model of LLM Activations", "authors": "Grace Luo, Jiahai Feng, Trevor Darrell, Alec Radford, Jacob Steinhardt", "year": 2026, "commentary": "", "link": "https://arxiv.org/pdf/2602.06964" } ] }, { "title": "Representation Learning", "description": "Interesting/illuminating/surprising articles and perspectives on representation learning.", "papers": [ { "title": "Deep Learning is Not So Mysterious or Different", "authors": "Andrew Gordon Wilson", "year": 2025, "commentary": "Super interesting and illuminating perspective explaining why supposedly deep-learning-unique phenomena like deep double descent, overparametrization, etc. can be explained using soft inductive biases and existing generalization frameworks. The references are a treasure trove!", "link": "https://openreview.net/pdf?id=42Au7FoD8F" }, { "title": "Cognitive Behaviors that Enable Self-Improving Reasoners", "authors": "Kanishk Gandhi, Ayush Chakravarthy, Anikait Singh, Nathan Lile, Noah D. Goodman", "year": 2025, "commentary": "Identifies four cognitive behaviors (verification, backtracking, subgoal setting, backward chaining) that predict whether a model can self-improve via RL. The key finding is striking: it's the presence of reasoning behaviors, not answer correctness, that matters. Models exposed to training data with proper reasoning patterns -- even incorrect answers -- matched the improvement of models that had these behaviors naturally. A useful framing for thinking about what 'reasoning' actually is in these systems.", "link": "https://arxiv.org/abs/2503.01307" }, { "title": "Harnessing the Universal Geometry of Embeddings", "authors": "Rishi Jha, Collin Zhang, Vitaly Shmatikov, John X. Morris", "year": 2025, "commentary": "Presents an unsupervised method for translating embeddings across different vector spaces using a universal latent representation based on the Platonic Representation Hypothesis. Shows that embeddings maintain similarity across models with different architectures and training data, while also revealing security vulnerabilities in vector databases.", "link": "https://arxiv.org/pdf/2505.12540" }, { "title": "Questioning Representational Optimism in Deep Learning: The Fractured Entangled Representation Hypothesis", "authors": "Akarsh Kumar, Jeff Clune, Joel Lehman, Kenneth O. Stanley", "year": 2025, "commentary": "", "link": "https://arxiv.org/pdf/2505.11581" }, { "title": "FIRE: Frobenius-Isometry Reinitialization for Balancing the Stability-Plasticity Tradeoff", "authors": "Isaac Han, Sangyeon Park, Seungwon Oh, Donghu Kim, Hojoon Lee, Kyung-Joong Kim", "year": 2026, "commentary": "", "link": "https://arxiv.org/pdf/2602.08040" } ] }, { "title": "Open-ended Modeling", "description": "Modeling unconstrained by a vanilla notion of raw 'ground truth' as supervisory signal.", "papers": [ { "title": "Unsupervised Elicitation of Language Models", "authors": "Jiaxin Wen, Zachary Ankner, Arushi Somani, Peter Hase, Samuel Marks, et al.", "year": 2025, "commentary": "Interesting way to automatically label datasets using mutual predictability and logical consistency.", "link": "https://arxiv.org/abs/2506.10139" } ] }, { "title": "Human-AI Interaction", "description": "Interesting HCI-oriented work exploring how we can interact with AI or with other humans mediated by AI.", "papers": [ { "title": "Embedding Comparator: Visualizing Differences in Global Structure and Local Neighborhoods via Small Multiples", "authors": "Angie Boggust, Brandon Carter, Arvind Satyanarayan", "year": 2022, "commentary": "A very useful interface for comparing differences in embeddings with interesting applications to model training development and the social sciences.", "link": "https://arxiv.org/abs/1912.04853" }, { "title": "Towards Bidirectional Human-AI Alignment: A Systematic Review for Clarifications, Framework, and Future Directions", "authors": "Hua Shen, Tiffany Knearem, Reshmi Ghosh, Kenan Alkiek, Kundan Krishna, et al.", "year": 2024, "commentary": "A recognition that not only must AI 'align to' human (values, behavior, knowledge, etc.) (-- whatever this means), but we also need to think about how humans might 'align' to AI by working with AI-structured systems. This paper recognizes social 'looping effects' brought about by AI and its behavior.", "link": "https://arxiv.org/abs/2406.09264" }, { "title": "AI Should Not Be An Imitation Game: Centaur Evaluations", "authors": "Andreas Haupt, Erik Brynjolfsson", "year": 2025, "commentary": "Advocates and proposes directions for systematic AI evalutions involving real human interaction.", "link": "https://openreview.net/pdf?id=LkdH35003E" }, { "title": "HCI for AGI", "authors": "Meredith Ringel Morris", "year": 2025, "commentary": "Useful outline of what HCI researchers can contribute to 'AGI'. It's not obvious (and people may fear that) interaction problems will be solved by AGI. Perhaps not?", "link": "https://dl.acm.org/doi/10.1145/3708815" }, { "title": "Code Shaping: Iterative Code Editing with Free-form AI-Interpreted Sketching", "authors": "Ryan Yen, Jian Zhao, Daniel Vogel", "year": 2025, "commentary": "Cool classic HCI-style exploration into a new interaction for code editing using visual sketching, interpreted by VLMs.", "link": "https://arxiv.org/abs/2502.03719" }, { "title": "Why Chatbots Are Not the Future", "authors": "Amelia Wattenberger", "year": 2025, "commentary": "Really nice argumentative piece on why we can build much better AI interfaces than chat interfaces.", "link": "https://wattenberger.com/thoughts/boo-chatbots" }, { "title": "Fish Eye for Text", "authors": "Amelia Wattenberger", "year": 2025, "commentary": "A beautifully designed webpage that illustrates a 'fisheye' principle for building knowledge interfaces that expose users to the 'peripheral' context.", "link": "https://wattenberger.com/thoughts/fish-eye" }, { "title": "CollabLLM: From Passive Responders to Active Collaborators", "authors": "Shirley Wu, Michel Galley, Baolin Peng, Hao Cheng, Gavin Li, Yao Dou, Weixin Cai, James Zou, Jure Leskovec, Jianfeng Gao", "year": 2025, "commentary": "Uses multiturn-aware rewards to optimize for long-term conversation quality rather than just immediate responses. Shows that LLMs trained this way actively seek clarification and offer suggestions, achieving 18.5% higher task performance and 46.3% better interactivity. Outstanding Paper Award at ICML 2025.", "link": "https://arxiv.org/abs/2502.00640" }, { "title": "Semantic Commit: Helping Users Update Intent Specifications for AI Memory at Scale", "authors": "Priyan Vaithilingam, Munyeong Kim, Frida-Cecilia Acosta-Parenteau, Daniel Lee, Amine Mhedhbi, Elena L. Glassman, Ian Arawjo", "year": 2025, "commentary": "", "link": "https://arxiv.org/abs/2504.09283" }, { "title": "Home-Cooked Software and Barefoot Developers", "authors": "Maggie Appleton", "year": 2024, "commentary": "Beautifully illustrated and thought-provoking talk about software that is 'home-cooked' -- not industrially produced like in a restaurant or factory, but made for a few people you care about to use. Introduces the concept of 'barefoot developers' (inspired by barefoot doctors in 1960s China) as a middle ground between professional developers and end users who don't want to code. Most people don't care about coding and don't want to, so we need barefoot developers. LLMs can help barefoot developers make more home-cooked software.", "link": "https://maggieappleton.com/home-cooked-software" }, { "title": "Antagonistic AI", "authors": "Alice Cai, Ian Arawjo, Elena L. Glassman", "year": 2024, "commentary": "One of my favorite papers and definitely a model for how to write good papers I think.", "link": "https://arxiv.org/abs/2402.07350" }, { "title": "Full-Stack Alignment: Co-Aligning AI and Institutions with Thick Models of Value", "authors": "Joe Edelman, Tan Zhi-Xuan, Ryan Lowe, Oliver Klingefjord, David Duvenaud, Jakob Foerster, Iason Gabriel, et al.", "year": 2025, "commentary": "The most exciting vision on AI alignment I've read in a while.", "link": "https://arxiv.org/pdf/2512.03399" }, { "title": "Who's in Charge? Disempowerment Patterns in Real-World LLM Usage", "authors": "Mrinank Sharma, Miles McCain, Raymond Douglas, David Duvenaud", "year": 2025, "commentary": "A nuanced operationalization of 'situational disempowerment' evaluated on a large corpus of Claude conversations. They also find that users tend to prefer model responses that lead to situational disempowerment -- people seem to like being told what to do, especially when they turn to AI for guidance -- which suggests that preference-based training may skew model behavior towards disempowering users.", "link": "https://arxiv.org/pdf/2601.19062" } ] }, { "title": "Philosophy and History of Math", "description": "Books and articles on philosophy of mathematics and history of mathematical ideas.", "papers": [ { "title": "Proofs and Refutations: The Logic of Mathematical Discovery", "authors": "Imre Lakatos", "year": 1976, "commentary": "Uses an extremely entertaining and well-focused example of understanding the 'V - E + F = 2' Euler characterization of polyhedra to illustrate how mathemematics develops -- dialectically between criticism (proof-analysis, refutation, counterexamples) and development (proof, lemma-incorporation, etc.). Entertainingly written as a Socratic conversation among a classroom of incredibly bright students.", "link": "https://dl1.cuni.cz/pluginfile.php/730446/mod_resource/content/2/Imre%20Lakatos%3B%20Proofs%20and%20Refutations.pdf" }, { "title": "A History of Abstract Algebra", "authors": "Isabelle Kleiner", "year": 2007, "commentary": "A comprehensive historical survey of the development of abstract algebra from its origins to modern times.", "link": "https://pyrkov-professor.ru/Portals/0/Mediateka/18%20vek/kleiner_i_a_history_of_abstract_algebra.pdf" }, { "title": "The Shape of Math To Come", "authors": "Alex Kontorovich", "year": 2025, "commentary": "Kontorovich reflects on how AI and formal verification systems like Lean are reshaping mathematical practice, intended for ICM 2026. What I find compelling is that the examples come from someone deeply embedded in both traditional research mathematics and these new technologies. The discussion of the 'Bitter Lesson' applied to mathematics is sobering, and the concrete examples of what formal verification looks like in practice are clarifying. A useful document for thinking about what 'doing math' might mean going forward.", "link": "https://arxiv.org/abs/2510.15924" }, { "title": "Neural Networks for Mathematical Discovery", "authors": "Geordie Williamson", "year": 2025, "commentary": "", "link": "https://www.youtube.com/watch?v=Uxr_HrbN1MQ" }, { "title": "One Man's View of Computer Science", "authors": "Richard Hamming", "year": 1968, "commentary": "Hamming's 1968 Turing Award lecture arguing for more of a practical, engineering flavor in computer science education.", "link": "https://dl.acm.org/doi/pdf/10.1145/1283920.1283923" } ] }, { "title": "Philosophy", "description": "Smaller articles/essays/books in contemporary-ish philosophy that I think are interesting or insightful.", "papers": [ { "title": "Why Isn't There More Progress in Philosophy?", "authors": "David J. Chalmers", "year": 2015, "commentary": "Asks an interesting question and gives some helpful directions to start thinking about what progress means in philosophy and how 'premise deniability' might be a relevant factor.", "link": "https://consc.net/papers/progress.pdf" }, { "title": "The Idea of Perfection", "authors": "Iris Murdoch", "year": 1964, "commentary": "From The Sovereignty of Good. Murdoch's influential essay on moral philosophy, arguing against the existentialist notion of radical choice and for the importance of attention and moral vision in ethical development.", "link": "" }, { "title": "Tools for Conviviality", "authors": "Ivan Illich", "year": 1973, "commentary": "An ambitious yet informed vision for what it would mean and cost for us to have *convivial* tools -- tools that we can make and shape our own lives for joy and purpose.", "link": "" }, { "title": "People need not only to obtain things; they need above all the freedom to make things among which they can live, to give shape to them according to their own tastes, and to put them to use in caring for and about others.", "authors": "Ivan Illich", "year": 1973, "commentary": "", "link": "" }, { "title": "Nietzsche and the Virtues of Mature Egoism", "authors": "Christine Swanton", "year": 2011, "commentary": "Swanton reads Nietzsche's 'immoralism' and 'egoism' as articulating virtues of the 'mature egoist' -- someone who has overcome immature egoism (ressentiment, self-deception, reactive self-assertion) in favor of genuine self-affirmation and creative engagement with the world. Setting aside questions of fidelity to Nietzsche's texts, I find this a compelling moral outlook: a rejection of both slavish self-denial and petty self-aggrandizement in favor of something like joyful, active, honest self-cultivation.", "link": "https://www.cambridge.org/core/books/abs/nietzsches-on-the-genealogy-of-morality/nietzsche-and-the-virtues-of-mature-egoism/7F4FAEEB715286707F9F5514400219A9" }, { "title": "Re-engineering Philosophy for Limited Beings", "authors": "William C. Wimsatt", "year": 2007, "commentary": "", "link": "" }, { "title": "The Professor of Parody", "authors": "Martha Nussbaum", "year": 1999, "commentary": "A critical analysis of Judith Butler's feminist theory and its influence on contemporary academic feminism.", "link": "https://perso.uclouvain.be/mylene.botbol/Recherche/GenreBioethique/Nussbaum_NRO.htm" }, { "title": "Intuitions of Compromise: Utilitarianism vs. Contractualism", "authors": "Jared Moore, Yejin Choi, Sydney Levine", "year": 2024, "commentary": "An interesting example of work in the more mathematical flavor of moral psychology / moral behavior in philosophy which also touches upon the AI space in a way that feels very responsible and respectful to the actual work that has been done in philosophy.", "link": "https://arxiv.org/abs/2410.05496" } ] }, { "title": "Miscellaneous", "description": "Interesting HCI-oriented work exploring how we can interact with AI or with other humans mediated by AI.", "papers": [ { "title": "The Economics of Maps", "authors": "Abhishek Nagaraj, Scott Stern", "year": 2020, "commentary": "An interesting analysis of economic issues in who uses and produces maps.", "link": "https://www.aeaweb.org/articles?id=10.1257/jep.34.1.196" }, { "title": "Self-reports are better measurement instruments than implicit measures", "authors": "Olivier Corneille, Bertram Gawronski", "year": 2024, "commentary": "Challenges the assumption that implicit measures are superior to self-reports in psychological research. Argues that self-reports demonstrate greater reliability, stronger predictive validity for both deliberate and spontaneous behaviors, and unmatched flexibility in exploring complex psychological constructs.", "link": "https://www.nature.com/articles/s44159-024-00376-z" }, { "title": "The Law of Leaky Abstractions", "authors": "Joel Spolsky", "year": 2002, "commentary": "", "link": "https://www.joelonsoftware.com/2002/11/11/the-law-of-leaky-abstractions/" }, { "title": "The Mythology Of Conscious AI", "authors": "Anil Seth", "year": 2026, "commentary": "There's the powerful idea that everything in AI is changing exponentially. Whether it's raw compute as indexed by Moore's Law, or the new capabilities available with each new iteration of the big tech foundation models, things surely are changing quickly. Exponential growth has the psychologically destabilizing property that what's ahead seems impossibly steep, and what's behind seems irrelevantly flat. Crucially, things seem this way wherever you are on the curve — that's what makes it exponential. Because of this, it's tempting to feel like we are always on the cusp of a major transition, and what could be more major than the creation of real artificial consciousness? But on an exponential curve, every point is an inflection point. ... The cartoon dreams of a silicon rapture, with its tropes of mind uploading, of disembodied eternal existence and of cloud-based reunions with other chosen ones, is a regression to the Cartesian soul. Computers, or more precisely computations, are, after all, immortal, and the sacrament of the algorithm promises a purist rationality, untainted by the body (despite plentiful evidence linking reason to emotion). But these are likely to be empty dreams, delivering not posthuman paradise but silicon oblivion. What really matters is not this kind of soul. Not any disembodied human-exceptionalist undying essence of you or of me. Perhaps what makes us us harks even further back, to Ancient Greece and to the plains of India, where our innermost essence arises as an inchoate feeling of just being alive — more breath than thought and more meat than machine. The sociologist Sherry Turkle once said that technology can make us forget what we know about life. It's about time we started to remember.", "link": "https://www.noemamag.com/the-mythology-of-conscious-ai/" }, { "title": "The Behavior of Ethicists", "authors": "Eric Schwitzgebel, Joshua Rust", "year": 2014, "commentary": "Both just funny and quite interesting work in experimental philosophy.", "link": "https://faculty.ucr.edu/~eschwitz/SchwitzPapers/BehEth-140123a.pdf" }, { "title": "Knowledge in the Head and in the World (Ch. 3, The Design of Everyday Things)", "authors": "Don Norman", "year": 1988, "commentary": "Recapitulates many ideas from psychology, philosophy, and anthropology through the lens of human tool-use and cognition, and does so in a way that is genuinely accessible and enjoyable to read. A personally rewarding chapter.", "link": "" } ] } ] }