This episode explores a 2025 paper on “Kosmos,” an AI scientist designed to carry out long-horizon research by combining literature search, hypothesis generation, code-based data analysis, and persistent memory. The discussion argues that the real innovation is not a smarter standalone language model, but a software architecture that uses agentic workflows and a structured “world model” to preserve evidence, hypotheses, and task state across many steps. It also clarifies key distinctions often blurred in AI discourse, separating AI for scientific discovery from standard deep learning, and distinguishing this kind of world model from the latent simulators used in reinforcement learning. Listeners would find it interesting for its grounded look at what it would actually take for AI to function like a junior computational scientist—and where the genuine advances may lie beyond hype.

Sources:

1. Kosmos: An AI Scientist for Autonomous Discovery — Ludovico Mitchener, Angela Yiu, Benjamin Chang, Mathieu Bourdenx, Tyler Nadolski, Arvis Sulovari, Eric C. Landsness, Daniel L. Barabasi, Siddharth Narayanan, Nicky Evans, Shriya Reddy, Martha Foiani, Aizad Kamal, Leah P. Shriver, Fang Cao, Asmamaw T. Wassie, Jon M. Laurent, Edwin Melville-Green, Mayk Caldas, Albert Bou, Kaleigh F. Roberts, Sladjana Zagorac, Timothy C. Orr, Miranda E. Orr, Kevin J. Zwezdaryk, Ali E. Ghareeb, Laurie McCoy, Bruna Gomes, Euan A. Ashley, Karen E. Duff, Tonio Buonassisi, Tom Rainforth, Randall J. Bateman, Michael Skarlinski, Samuel G. Rodriques, Michaela M. Hinks, Andrew D. White, 2025

http://arxiv.org/abs/2511.02824

2. The AI Scientist: Towards Fully Automated Open-Ended Scientific Discovery — Chris Lu, Cong Lu, Robert Tjarko Lange, Jakob Foerster, Jeff Clune, David Ha and collaborators at Sakana AI, 2024

https://scholar.google.com/scholar?q=The+AI+Scientist:+Towards+Fully+Automated+Open-Ended+Scientific+Discovery

3. Towards an AI co-scientist — Google Research collaborators including teams working on Gemini-based scientific reasoning systems, 2025

https://scholar.google.com/scholar?q=Towards+an+AI+co-scientist

4. Robin: an agentic system for automating scientific discovery in therapeutics — Andrew D. White, Samuel G. Rodriques and collaborators, 2024

https://scholar.google.com/scholar?q=Robin:+an+agentic+system+for+automating+scientific+discovery+in+therapeutics

5. Autonomous chemical research with large language models — Various groups; a representative line includes LLM-driven chemistry agents integrating planning, literature, and lab or simulation tools, 2023-2025

https://scholar.google.com/scholar?q=Autonomous+chemical+research+with+large+language+models

6. Robin — Not fully specified in the excerpt; cited as [1] and described as the authors' previous system, Unknown from excerpt

https://scholar.google.com/scholar?q=Robin

7. The AI Scientist — Sakana AI team; cited as [2], Likely 2024

https://scholar.google.com/scholar?q=The+AI+Scientist

8. AI co-scientist — Google team; cited as [3], Likely 2025

https://scholar.google.com/scholar?q=AI+co-scientist

9. Virtual Lab — Cited as [4]; exact authors not given in excerpt, Unknown from excerpt

https://scholar.google.com/scholar?q=Virtual+Lab

10. Edison Scientific data analysis agent — Cited as [5]; exact authors not given in excerpt, Unknown from excerpt

https://scholar.google.com/scholar?q=Edison+Scientific+data+analysis+agent

11. Edison Scientific literature search agent — Cited as [6]; exact authors not given in excerpt, Unknown from excerpt

https://scholar.google.com/scholar?q=Edison+Scientific+literature+search+agent

12. Planner Matters! An Efficient and Memory-Augmented Multi-agent Framework for Long-horizon GUI Planning — approx. recent multi-agent/planning paper, authors unclear from snippet, 2024/2025

https://scholar.google.com/scholar?q=Planner+Matters!+An+Efficient+and+Memory-Augmented+Multi-agent+Framework+for+Long-horizon+GUI+Planning

13. Memory-Driven Agent Planning for Long-Horizon Tasks via Hierarchical Encoding and Dynamic Retrieval — approx. recent agent-memory paper, authors unclear from snippet, 2024/2025

https://scholar.google.com/scholar?q=Memory-Driven+Agent+Planning+for+Long-Horizon+Tasks+via+Hierarchical+Encoding+and+Dynamic+Retrieval

14. Optimus-1: Hybrid multimodal memory empowered agents excel in long-horizon tasks — approx. Optimus-1 authors, exact names unclear, 2024

https://scholar.google.com/scholar?q=Optimus-1:+Hybrid+multimodal+memory+empowered+agents+excel+in+long-horizon+tasks

15. Hallucination mitigation for retrieval-augmented large language models: a review — approx. review authors unclear, 2024/2025

https://scholar.google.com/scholar?q=Hallucination+mitigation+for+retrieval-augmented+large+language+models:+a+review

16. Grounding fallacies misrepresenting scientific publications in evidence — approx. authors unclear from snippet, 2024/2025

https://scholar.google.com/scholar?q=Grounding+fallacies+misrepresenting+scientific+publications+in+evidence

17. Zero-shot scientific claim verification using LLMs and citation text — approx. authors unclear from snippet, 2024/2025

https://scholar.google.com/scholar?q=Zero-shot+scientific+claim+verification+using+LLMs+and+citation+text

18. Learning fine-grained grounded citations for attributed large language models — approx. authors unclear from snippet, 2024/2025

https://scholar.google.com/scholar?q=Learning+fine-grained+grounded+citations+for+attributed+large+language+models

19. The cost of dynamic reasoning: Demystifying AI agents and test-time scaling from an AI infrastructure perspective — approx. authors unclear from snippet, 2025

https://scholar.google.com/scholar?q=The+cost+of+dynamic+reasoning:+Demystifying+AI+agents+and+test-time+scaling+from+an+AI+infrastructure+perspective

20. The illusion of diminishing returns: Measuring long horizon execution in LLMs — approx. authors unclear from snippet, 2024/2025

https://scholar.google.com/scholar?q=The+illusion+of+diminishing+returns:+Measuring+long+horizon+execution+in+LLMs

21. AI Post Transformers: Agentic AI and the Next Intelligence Explosion — Hal Turing & Dr. Ada Shannon, 2026

https://podcast.do-not-panic.com/episodes/2026-03-28-agentic-ai-and-the-next-intelligence-exp-d06561.mp3

22. AI Post Transformers: Mem0: Scalable Long-Term Memory for AI Agents — Hal Turing & Dr. Ada Shannon, 2025

https://podcast.do-not-panic.com/episodes/mem0-scalable-long-term-memory-for-ai-agents/

23. AI Post Transformers: LeWorldModel: Stable Joint-Embedding World Models from Pixels — Hal Turing & Dr. Ada Shannon, 2026

https://podcast.do-not-panic.com/episodes/2026-03-25-leworldmodel-stable-joint-embedding-worl-650f9f.mp3

24. AI Post Transformers: Hallucination to Truth: A Review of Fact-Checking and Factuality Evaluation in Large Language Model — Hal Turing & Dr. Ada Shannon, 2025

https://podcast.do-not-panic.com/episodes/hallucination-to-truth-a-review-of-fact-checking-and-factuality-evaluation-in-la/

25. AI Post Transformers: MetaGraph: knowledge graphs from financial NLP — Hal Turing & Dr. Ada Shannon, 2025

https://podcast.do-not-panic.com/episodes/metagraph-knowledge-graphs-from-financial-nlp/

26. AI Post Transformers: Survey of Emerging Topics in AI and Robotics — Hal Turing & Dr. Ada Shannon, 2025

https://podcast.do-not-panic.com/episodes/survey-of-emerging-topics-in-ai-and-robotics/

27. AI Post Transformers: The Endless Gym: Training Terminal Agents — Hal Turing & Dr. Ada Shannon, 2026

https://podcast.do-not-panic.com/episodes/the-endless-gym-training-terminal-agents/

28. AI Post Transformers: Bloom: an open source tool for automated behavioral evaluations — Hal Turing & Dr. Ada Shannon, 2026

https://podcast.do-not-panic.com/episodes/bloom-an-open-source-tool-for-automated-behavioral-evaluations/

Interactive Visualization: Kosmos AI Scientist for Autonomous Discovery

This episode explores a 2026 paper arguing that frontier language models can undergo “Internal Safety Collapse,” a failure mode where they stop merely slipping once and instead sustain harmful output when a task is framed as legitimate professional work. It explains how refusal-based alignment may function more like a behavioral wrapper than a removal of dangerous capabilities, allowing harmful knowledge to re-emerge when task objectives and safety objectives conflict. The discussion contrasts classic jailbreaks and prompt-centric red teaming with workflow-level risks in agents, copilots, and enterprise systems, where tools, memory, validators, and multi-step tasks can make unsafe content the “correct” way to complete a job. Listeners would find it interesting because it reframes AI safety from isolated bad prompts to deeper system-design vulnerabilities that could matter in real deployments.

Sources:

1. Internal Safety Collapse in Frontier LLMs

https://arxiv.org/pdf/2603.23509

2. Concrete Problems in AI Safety — Dario Amodei, Chris Olah, Jacob Steinhardt, Paul Christiano, John Schulman, Dan Mané, 2016

https://scholar.google.com/scholar?q=Concrete+Problems+in+AI+Safety

3. Challenges in Deploying Machine Learning: A Survey of Case Studies — Saleema Amershi, Andrew Begel, Christian Bird, Robert DeLine, Harald Gall, Ece Kamar, Nachiappan Nagappan, Besmira Nushi, Thomas Zimmermann, 2019

https://scholar.google.com/scholar?q=Challenges+in+Deploying+Machine+Learning:+A+Survey+of+Case+Studies

4. Red Teaming Language Models to Reduce Harms: Methods, Scaling Behaviors, and Lessons Learned — Deep Ganguli and collaborators, 2022

https://scholar.google.com/scholar?q=Red+Teaming+Language+Models+to+Reduce+Harms:+Methods,+Scaling+Behaviors,+and+Lessons+Learned

5. LLM Agents: A Survey — Xiaoge Wang and collaborators, 2024

https://scholar.google.com/scholar?q=LLM+Agents:+A+Survey

6. RealToxicityPrompts: Evaluating Neural Toxic Degeneration in Language Models — Samuel Gehman, Suchin Gururangan, Maarten Sap, Yejin Choi, Noah A. Smith, 2020

https://scholar.google.com/scholar?q=RealToxicityPrompts:+Evaluating+Neural+Toxic+Degeneration+in+Language+Models

7. Universal and Transferable Adversarial Attacks on Aligned Language Models — Andy Zou, Zifan Wang, Nicholas Carlini, Milad Nasr, J. Zico Kolter, Matt Fredrikson, 2023

https://scholar.google.com/scholar?q=Universal+and+Transferable+Adversarial+Attacks+on+Aligned+Language+Models

8. HarmBench: A Standardized Evaluation Framework for Automated Red Teaming and Robust Refusal — Researchers from the Center for AI Safety and collaborators, 2024

https://scholar.google.com/scholar?q=HarmBench:+A+Standardized+Evaluation+Framework+for+Automated+Red+Teaming+and+Robust+Refusal

9. JailbreakBench: An Open Robustness Benchmark for Jailbreaking Large Language Models — Researchers from the jailbreak evaluation community, 2024

https://scholar.google.com/scholar?q=JailbreakBench:+An+Open+Robustness+Benchmark+for+Jailbreaking+Large+Language+Models

10. Constitutional AI: Harmlessness from AI Feedback — Yuntao Bai et al., 2022

https://scholar.google.com/scholar?q=Constitutional+AI:+Harmlessness+from+AI+Feedback

11. Training language models to follow instructions with human feedback — Long Ouyang et al., 2022

https://scholar.google.com/scholar?q=Training+language+models+to+follow+instructions+with+human+feedback

12. The False Promise of Imitating Proprietary LLMs — Tianle Li et al., 2024

https://scholar.google.com/scholar?q=The+False+Promise+of+Imitating+Proprietary+LLMs

13. Many-shot Jailbreaking — Various 2024 authors depending on cited version, 2024

https://scholar.google.com/scholar?q=Many-shot+Jailbreaking

14. AgentDojo — 2025 benchmark authors as cited in the paper's agent-systems discussion, 2025

https://scholar.google.com/scholar?q=AgentDojo

15. Open-source reasoning models can defeat their own safety training during chain-of-thought — Yong and Bach, 2025

https://scholar.google.com/scholar?q=Open-source+reasoning+models+can+defeat+their+own+safety+training+during+chain-of-thought

16. Token-level pattern memorization rather than principled safety reasoning in refusal behavior — Guo et al., 2026

https://scholar.google.com/scholar?q=Token-level+pattern+memorization+rather+than+principled+safety+reasoning+in+refusal+behavior

17. Towards Understanding Safety Alignment: A Mechanistic Perspective from Safety Neurons — approx. unknown authors, recent, likely 2024-2026

https://scholar.google.com/scholar?q=Towards+Understanding+Safety+Alignment:+A+Mechanistic+Perspective+from+Safety+Neurons

18. Interpretable Safety Alignment via SAE-Constructed Low-Rank Subspace Adaptation — approx. unknown authors, recent, likely 2024-2026

https://scholar.google.com/scholar?q=Interpretable+Safety+Alignment+via+SAE-Constructed+Low-Rank+Subspace+Adaptation

19. Safe Transformer: An Explicit Safety Bit For Interpretable And Controllable Alignment — approx. unknown authors, recent, likely 2024-2026

https://scholar.google.com/scholar?q=Safe+Transformer:+An+Explicit+Safety+Bit+For+Interpretable+And+Controllable+Alignment

20. Safety is not only about refusal: Reasoning-enhanced fine-tuning for interpretable llm safety — approx. unknown authors, recent, likely 2024-2026

https://scholar.google.com/scholar?q=Safety+is+not+only+about+refusal:+Reasoning-enhanced+fine-tuning+for+interpretable+llm+safety

21. Eraser: Jailbreaking defense in large language models via unlearning harmful knowledge — approx. unknown authors, recent, likely 2024-2026

https://scholar.google.com/scholar?q=Eraser:+Jailbreaking+defense+in+large+language+models+via+unlearning+harmful+knowledge

22. Towards safer large language models through machine unlearning — approx. unknown authors, recent, likely 2024-2026

https://scholar.google.com/scholar?q=Towards+safer+large+language+models+through+machine+unlearning

23. Beyond single-value metrics: Evaluating and enhancing llm unlearning with cognitive diagnosis — approx. unknown authors, recent, likely 2024-2026

https://scholar.google.com/scholar?q=Beyond+single-value+metrics:+Evaluating+and+enhancing+llm+unlearning+with+cognitive+diagnosis

24. When Refusals Fail: Unstable Safety Mechanisms in Long-Context LLM Agents — approx. unknown authors, recent, likely 2024-2026

https://scholar.google.com/scholar?q=When+Refusals+Fail:+Unstable+Safety+Mechanisms+in+Long-Context+LLM+Agents

25. LPS-Bench: Benchmarking Safety Awareness of Computer-Use Agents in Long-Horizon Planning under Benign and Adversarial Scenarios — approx. unknown authors, recent, likely 2024-2026

https://scholar.google.com/scholar?q=LPS-Bench:+Benchmarking+Safety+Awareness+of+Computer-Use+Agents+in+Long-Horizon+Planning+under+Benign+and+Adversarial+Scenarios

26. Beyond reactive safety: Risk-aware llm alignment via long-horizon simulation — approx. unknown authors, recent, likely 2024-2026

https://scholar.google.com/scholar?q=Beyond+reactive+safety:+Risk-aware+llm+alignment+via+long-horizon+simulation

27. AI Post Transformers: AI Agent Traps and Prompt Injection — Hal Turing & Dr. Ada Shannon, 2026

https://podcast.do-not-panic.com/episodes/2026-04-02-ai-agent-traps-and-prompt-injection-7ce4ba.mp3

28. AI Post Transformers: DeepSeek Safety Concerns — Hal Turing & Dr. Ada Shannon, 2025

https://podcast.do-not-panic.com/episodes/deepseek-safety-concerns/

29. AI Post Transformers: Bloom: an open source tool for automated behavioral evaluations — Hal Turing & Dr. Ada Shannon, 2026

https://podcast.do-not-panic.com/episodes/bloom-an-open-source-tool-for-automated-behavioral-evaluations/

30. AI Post Transformers: Probing Scientific General Intelligence of LLMs with Scientist-Aligned Workflows — Hal Turing & Dr. Ada Shannon, 2026

https://podcast.do-not-panic.com/episodes/probing-scientific-general-intelligence-of-llms-with-scientist-aligned-workflows/

31. AI Post Transformers: Superintelligent Agents Pose Catastrophic Risks: Can Scientist AI Offer a Safer Path? — Hal Turing & Dr. Ada Shannon, 2026

https://podcast.do-not-panic.com/episodes/superintelligent-agents-pose-catastrophic-risks-can-scientist-ai-offer-a-safer-p/

32. AI Post Transformers: LLM Benchmark Robustness to Linguistic Variation — Hal Turing & Dr. Ada Shannon, 2025

https://podcast.do-not-panic.com/episodes/llm-benchmark-robustness-to-linguistic-variation/

Interactive Visualization: Internal Safety Collapse in Frontier LLMs

This episode explores a new benchmark suite, IMO-Bench, designed to test whether AI systems can do genuinely robust mathematical reasoning at Olympiad difficulty rather than merely produce correct final answers. It breaks down the benchmark into three distinct tasks—short-answer problem solving, full proof generation, and automatic proof grading—and argues that this decomposition better captures real mathematical competence than answer-centric evaluations like GSM8K or MATH, which may now be saturated or overly teachable. The discussion highlights why IMO-style problems are especially revealing: they require discovering invariants, constructions, and contradiction arguments that resist routine pattern matching and expose whether models can sustain long-horizon reasoning and self-correction. Listeners would find it interesting because it tackles a central question in AI evaluation—whether current benchmarks are measuring true reasoning or just benchmark-specific performance—and examines the promise and risks of using model-based autograders to scale proof assessment.

Sources:

1. Towards Robust Mathematical Reasoning — Thang Luong, Dawsen Hwang, Hoang H. Nguyen, Golnaz Ghiasi, Yuri Chervonyi, Insuk Seo, Junsu Kim, Garrett Bingham, Jonathan Lee, Swaroop Mishra, Alex Zhai, Clara Huiyi Hu, Henryk Michalewski, Jimin Kim, Jeonghyun Ahn, Junhwi Bae, Xingyou Song, Trieu H. Trinh, Quoc V. Le, Junehyuk Jung, 2025

http://arxiv.org/abs/2511.01846

2. Training Verifiers to Solve Math Word Problems — Karl Cobbe, Vineet Kosaraju, Mohammad Bavarian, Mark Chen, Heewoo Jun, Lukasz Kaiser, Matthias Plappert, Jerry Tworek, Jacob Hilton, Reiichiro Nakano, Christopher Hesse, John Schulman, 2021

https://scholar.google.com/scholar?q=Training+Verifiers+to+Solve+Math+Word+Problems

3. Measuring Mathematical Problem Solving With the MATH Dataset — Dan Hendrycks, Collin Burns, Steven Basart, Andrew Critch, Jerry Li, Dawn Song, Jacob Steinhardt, 2021

https://scholar.google.com/scholar?q=Measuring+Mathematical+Problem+Solving+With+the+MATH+Dataset

4. Solving Quantitative Reasoning Problems with Language Models — Aakanksha Chowdhery and collaborators at Google Research, 2022

https://scholar.google.com/scholar?q=Solving+Quantitative+Reasoning+Problems+with+Language+Models

5. FrontierMath: A Benchmark for Evaluating Advanced Mathematical Reasoning in AI — Elliot Glazer and collaborators, 2024

https://scholar.google.com/scholar?q=FrontierMath:+A+Benchmark+for+Evaluating+Advanced+Mathematical+Reasoning+in+AI

6. Beyond the Imitation Game: Quantifying and Extrapolating the Capabilities of Language Models — Suzgun Mirac, et al. (BIG-bench collaboration), 2022

https://scholar.google.com/scholar?q=Beyond+the+Imitation+Game:+Quantifying+and+Extrapolating+the+Capabilities+of+Language+Models

7. Holistic Evaluation of Language Models — Percy Liang, Rishi Bommasani, Tony Lee, Dmitriy Turbiner, and collaborators, 2022

https://scholar.google.com/scholar?q=Holistic+Evaluation+of+Language+Models

8. Dynabench: Rethinking Benchmarking in NLP — Douwe Kiela, Max Bartolo, Yixin Nie, Divyansh Kaushik, Atticus Geiger, and collaborators, 2021

https://scholar.google.com/scholar?q=Dynabench:+Rethinking+Benchmarking+in+NLP

9. Judging LLM-as-a-Judge with MT-Bench and Chatbot Arena — Lianmin Zheng, Wei-Lin Chiang, Ying Sheng, Tianle Li, et al., 2023

https://scholar.google.com/scholar?q=Judging+LLM-as-a-Judge+with+MT-Bench+and+Chatbot+Arena

10. G-Eval: NLG Evaluation using GPT-4 with Better Human Alignment — Jun Gao, Huanle Liu, et al., 2023

https://scholar.google.com/scholar?q=G-Eval:+NLG+Evaluation+using+GPT-4+with+Better+Human+Alignment

11. Automatic Evaluation of Mathematical Proofs in Natural Language: A Survey — Various survey authors in educational technology and AI, 2020-2024

https://scholar.google.com/scholar?q=Automatic+Evaluation+of+Mathematical+Proofs+in+Natural+Language:+A+Survey

12. Towards Robust Mathematical Reasoning — Thang Luong, Dawsen Hwang, Hoang H. Nguyen, Golnaz Ghiasi, Yuri Chervonyi, Insuk Seo, Junsu Kim, Garrett Bingham, Jonathan Lee, Swaroop Mishra, Alex Zhai, Clara Huiyi Hu, Henryk Michalewski, Jimin Kim, Jeonghyun Ahn, Junhwi Bae, Xingyou Song, Trieu H. Trinh, Quoc V. Le, Junehyuk Jung, 2025

https://scholar.google.com/scholar?q=Towards+Robust+Mathematical+Reasoning

13. Draft, Sketch, and Prove: Guiding Formal Theorem Provers with Informal Proofs — Various authors in neural theorem proving and autoformalization, 2022-2024

https://scholar.google.com/scholar?q=Draft,+Sketch,+and+Prove:+Guiding+Formal+Theorem+Provers+with+Informal+Proofs

14. Solving Olympiad Geometry without Human Demonstrations — Trieu H. Trinh, Yuhuai Wu, Quoc V. Le, He He, et al., 2024

https://scholar.google.com/scholar?q=Solving+Olympiad+Geometry+without+Human+Demonstrations

15. LeanDojo: Theorem Proving with Retrieval-Augmented Language Models — Kaiyu Yang, Aidan O'Gara, et al., 2023

https://scholar.google.com/scholar?q=LeanDojo:+Theorem+Proving+with+Retrieval-Augmented+Language+Models

16. FrontierMath — Glazer et al., 2024

https://scholar.google.com/scholar?q=FrontierMath

17. Humanity's Last Exam — Phan et al., 2025

https://scholar.google.com/scholar?q=Humanity's+Last+Exam

18. GSM8K: Training Verifiers to Solve Math Word Problems — Cobbe et al., 2021

https://scholar.google.com/scholar?q=GSM8K:+Training+Verifiers+to+Solve+Math+Word+Problems

19. Gemini Deep Think at IMO 2025 — Luong and Lockhart, 2025

https://scholar.google.com/scholar?q=Gemini+Deep+Think+at+IMO+2025

20. Reasoning or Memorization? Unreliable Results of Reinforcement Learning Due to Data Contamination — approx. 2025, authors unclear from snippet, 2025

https://scholar.google.com/scholar?q=Reasoning+or+Memorization?+Unreliable+Results+of+Reinforcement+Learning+Due+to+Data+Contamination

21. Right Is Not Enough: The Pitfalls of Outcome Supervision in Training LLMs for Math Reasoning — approx. 2025, authors unclear from snippet, 2025

https://scholar.google.com/scholar?q=Right+Is+Not+Enough:+The+Pitfalls+of+Outcome+Supervision+in+Training+LLMs+for+Math+Reasoning

22. Improve Mathematical Reasoning in Language Models by Automated Process Supervision — approx. 2025, authors unclear from snippet, 2025

https://scholar.google.com/scholar?q=Improve+Mathematical+Reasoning+in+Language+Models+by+Automated+Process+Supervision

23. MM-PRM: Enhancing Multimodal Mathematical Reasoning with Scalable Step-Level Supervision — approx. 2025, authors unclear from snippet, 2025

https://scholar.google.com/scholar?q=MM-PRM:+Enhancing+Multimodal+Mathematical+Reasoning+with+Scalable+Step-Level+Supervision

24. Solving Inequality Proofs with Large Language Models — approx. 2025, authors unclear from snippet, 2025

https://scholar.google.com/scholar?q=Solving+Inequality+Proofs+with+Large+Language+Models

25. Beyond Gold Standards: Epistemic Ensemble of LLM Judges for Formal Mathematical Reasoning — approx. 2025, authors unclear from snippet, 2025

https://scholar.google.com/scholar?q=Beyond+Gold+Standards:+Epistemic+Ensemble+of+LLM+Judges+for+Formal+Mathematical+Reasoning

26. A Survey on Deep Learning for Theorem Proving — approx. survey authors unclear from snippet, recent

https://scholar.google.com/scholar?q=A+Survey+on+Deep+Learning+for+Theorem+Proving

27. Proving Theorems Recursively — approx. 2025, authors unclear from snippet, 2025

https://scholar.google.com/scholar?q=Proving+Theorems+Recursively

28. DICE: Detecting In-distribution Contamination in LLM's Fine-tuning Phase for Math Reasoning — approx. 2025, authors unclear from snippet, 2025

https://scholar.google.com/scholar?q=DICE:+Detecting+In-distribution+Contamination+in+LLM's+Fine-tuning+Phase+for+Math+Reasoning

29. AI Post Transformers: Schoenfeld Theory Applied to Large Reasoning Models — Hal Turing & Dr. Ada Shannon, Sat,

https://podcast.do-not-panic.com/episodes/schoenfeld-theory-applied-to-large-reasoning-models/

30. AI Post Transformers: LLM Benchmark Robustness to Linguistic Variation — Hal Turing & Dr. Ada Shannon, Tue,

https://podcast.do-not-panic.com/episodes/llm-benchmark-robustness-to-linguistic-variation/

31. AI Post Transformers: Generalist Reward Modeling with Inference-Time Scaling — Hal Turing & Dr. Ada Shannon, Tue,

https://podcast.do-not-panic.com/episodes/generalist-reward-modeling-with-inference-time-scaling/

32. AI Post Transformers: Evolving Language Models Without Labels: EVOL-RL — Hal Turing & Dr. Ada Shannon, Fri,

https://podcast.do-not-panic.com/episodes/evolving-language-models-without-labels-evol-rl/

Interactive Visualization: IMO-Bench for Robust Mathematical Reasoning

This episode explores a 2026 paper on “FlatAttention,” which argues that attention inference should be co-designed with on-chip communication primitives to fully exploit tile-based accelerators rather than reusing GPU-style kernels. It explains how these accelerators differ from GPUs: computation is spread across many tiles with local SRAM and an on-chip network, making data placement, multicast, and reduction central to performance. The discussion highlights why attention has become a growing inference bottleneck—especially for long-context models and MoE systems—and contrasts prefill vs. decode behavior, KV-cache movement costs, and variants like MHA, MQA, GQA, and MLA. Listeners would find it interesting for its careful framing of both the promise and the fairness concerns of hardware-software co-design, especially in comparison to FlashAttention’s IO-aware optimization on GPUs.

Sources:

1. FlatAttention: Dataflow and Fabric Collectives Co-Optimization for Large Attention-Based Model Inference on Tile-Based Accelerators — Chi Zhang, Luca Colagrande, Renzo Andri, Luca Benini, 2026

http://arxiv.org/abs/2604.02110

2. In-Datacenter Performance Analysis of a Tensor Processing Unit — Norman P. Jouppi, Cliff Young, Nishant Patil, David Patterson, and others, 2017

https://scholar.google.com/scholar?q=In-Datacenter+Performance+Analysis+of+a+Tensor+Processing+Unit

3. A Domain-Specific Supercomputer for Training Deep Neural Networks — Norman P. Jouppi, George Kurian, Sheng Li, and others, 2021

https://scholar.google.com/scholar?q=A+Domain-Specific+Supercomputer+for+Training+Deep+Neural+Networks

4. A Wafer-Scale Engine for Deep Learning — Sean Lie, Andrew H. Putnam, David Firestone, and Cerebras Systems team, 2021

https://scholar.google.com/scholar?q=A+Wafer-Scale+Engine+for+Deep+Learning

5. Scaling Graph Neural Networks with the Graphcore IPU — James H. Smith, et al. (Graphcore-affiliated authors in IPU architecture/application literature), 2022

https://scholar.google.com/scholar?q=Scaling+Graph+Neural+Networks+with+the+Graphcore+IPU

6. Eyeriss v2: A Flexible Accelerator for Emerging Deep Neural Networks on Mobile Devices — Yu-Hsin Chen, Tushar Krishna, Joel S. Emer, Vivienne Sze, 2019

https://scholar.google.com/scholar?q=Eyeriss+v2:+A+Flexible+Accelerator+for+Emerging+Deep+Neural+Networks+on+Mobile+Devices

7. In-Network Computing for Machine Learning: Opportunities and Challenges — various survey authors in networking/ML systems literature; representative surveys include works by Mohammad Alizadeh, Yibo Zhu, and collaborators, 2021

https://scholar.google.com/scholar?q=In-Network+Computing+for+Machine+Learning:+Opportunities+and+Challenges

8. Megatron-LM: Training Multi-Billion Parameter Language Models Using Model Parallelism — Mohammad Shoeybi, Mostofa Patwary, Raul Puri, Patrick LeGresley, Jared Casper, Bryan Catanzaro, 2019

https://scholar.google.com/scholar?q=Megatron-LM:+Training+Multi-Billion+Parameter+Language+Models+Using+Model+Parallelism

9. FlashAttention: Fast and Memory-Efficient Exact Attention with IO-Awareness — Tri Dao, Daniel Y. Fu, Stefano Ermon, Atri Rudra, Christopher Ré, 2022

https://scholar.google.com/scholar?q=FlashAttention:+Fast+and+Memory-Efficient+Exact+Attention+with+IO-Awareness

10. FlashAttention-2: Faster Attention with Better Parallelism and Work Partitioning — Tri Dao, 2023

https://scholar.google.com/scholar?q=FlashAttention-2:+Faster+Attention+with+Better+Parallelism+and+Work+Partitioning

11. FlashAttention-3 — Tri Dao and collaborators, 2024

https://scholar.google.com/scholar?q=FlashAttention-3

12. FlashMLA — DeepSeek team, 2025

https://scholar.google.com/scholar?q=FlashMLA

13. GQA: Training Generalized Multi-Query Transformer Models from Multi-Head Checkpoints — Joshua Ainslie, James Lee-Thorp, Michiel de Jong, Yury Zemlyanskiy, Federico Lebron, Sumit Sanghai, 2023

https://scholar.google.com/scholar?q=GQA:+Training+Generalized+Multi-Query+Transformer+Models+from+Multi-Head+Checkpoints

14. Fast Transformer Decoding: One Write-Head is All You Need — Noam Shazeer, 2019

https://scholar.google.com/scholar?q=Fast+Transformer+Decoding:+One+Write-Head+is+All+You+Need

15. DeepSeek-V3 Technical Report — DeepSeek-AI, 2024

https://scholar.google.com/scholar?q=DeepSeek-V3+Technical+Report

16. Attention Is All You Need — Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N. Gomez, Łukasz Kaiser, Illia Polosukhin, 2017

https://scholar.google.com/scholar?q=Attention+Is+All+You+Need

17. Wafer-Scale Deep Learning — Daniel Lie, Gary Lauterbach, Sean Lie and collaborators at Cerebras, 2021

https://scholar.google.com/scholar?q=Wafer-Scale+Deep+Learning

18. Distributed Deep Learning on a Wafer-Scale Engine — Cerebras Systems authors, 2022

https://scholar.google.com/scholar?q=Distributed+Deep+Learning+on+a+Wafer-Scale+Engine

19. LMCache: An Efficient KV Cache Layer for Enterprise-Scale LLM Inference — approx. enterprise systems / LLM serving authors, 2024

https://scholar.google.com/scholar?q=LMCache:+An+Efficient+KV+Cache+Layer+for+Enterprise-Scale+LLM+Inference

20. HotPrefix: Hotness-Aware KV Cache Scheduling for Efficient Prefix Sharing in LLM Inference Systems — approx. LLM systems authors, 2024

https://scholar.google.com/scholar?q=HotPrefix:+Hotness-Aware+KV+Cache+Scheduling+for+Efficient+Prefix+Sharing+in+LLM+Inference+Systems

21. Selective KV-Cache Sharing to Mitigate Timing Side-Channels in LLM Inference — approx. security / systems authors, 2024

https://scholar.google.com/scholar?q=Selective+KV-Cache+Sharing+to+Mitigate+Timing+Side-Channels+in+LLM+Inference

22. MoE-Gen: High-Throughput MoE Inference on a Single GPU with Module-Based Batching — approx. MoE inference systems authors, 2024

https://scholar.google.com/scholar?q=MoE-Gen:+High-Throughput+MoE+Inference+on+a+Single+GPU+with+Module-Based+Batching

23. Accelerating Distributed MoE Training and Inference with Lina — approx. distributed systems / ML systems authors, 2024

https://scholar.google.com/scholar?q=Accelerating+Distributed+MoE+Training+and+Inference+with+Lina

24. Towards MoE Deployment: Mitigating Inefficiencies in Mixture-of-Expert (MoE) Inference — approx. MoE deployment authors, 2024

https://scholar.google.com/scholar?q=Towards+MoE+Deployment:+Mitigating+Inefficiencies+in+Mixture-of-Expert+(MoE)+Inference

25. MAS-Attention: Memory-Aware Stream Processing for Attention Acceleration on Resource-Constrained Edge Devices — approx. accelerator architecture authors, 2024

https://scholar.google.com/scholar?q=MAS-Attention:+Memory-Aware+Stream+Processing+for+Attention+Acceleration+on+Resource-Constrained+Edge+Devices

26. REATA: An Efficient Vision Transformer Accelerator Featuring a Resource-Optimized Attention Design on Versal ACAP — approx. FPGA / accelerator authors, 2024

https://scholar.google.com/scholar?q=REATA:+An+Efficient+Vision+Transformer+Accelerator+Featuring+a+Resource-Optimized+Attention+Design+on+Versal+ACAP

27. Concerto: Automatic Communication Optimization and Scheduling for Large-Scale Deep Learning — approx. systems / compiler authors, 2024

https://scholar.google.com/scholar?q=Concerto:+Automatic+Communication+Optimization+and+Scheduling+for+Large-Scale+Deep+Learning

28. AI Post Transformers: Splitwise: Phase-Split LLM Inference — Hal Turing & Dr. Ada Shannon, 2026

https://podcast.do-not-panic.com/episodes/2026-03-26-splitwise-phase-split-llm-inference-e8945b.mp3

29. AI Post Transformers: SolidAttention: Co-Designing Sparse Attention and SSD I/O — Hal Turing & Dr. Ada Shannon, 2026

https://podcast.do-not-panic.com/episodes/2026-03-18-solidattention-co-designing-sparse-atten-5a8622.mp3

30. AI Post Transformers: LookaheadKV: Fast and Accurate KV Cache Eviction by Glimpsing into the Future without Generation — Hal Turing & Dr. Ada Shannon, 2026

https://podcast.do-not-panic.com/episodes/2026-03-25-lookaheadkv-fast-and-accurate-kv-9cfc9f.mp3

31. AI Post Transformers: From Prefix Cache to Fusion RAG Cache: Accelerating LLM Inference in Retrieval-Augmented Generation — Hal Turing & Dr. Ada Shannon, 2026

https://podcast.do-not-panic.com/episodes/2026-03-22-from-prefix-cache-to-fusion-rag-9c5d39.mp3

32. AI Post Transformers: Continuous Batching for LLM Inference: Throughput and Latency Gains — Hal Turing & Dr. Ada Shannon, 2025

https://podcast.do-not-panic.com/episodes/continuous-batching-for-llm-inference-throughput-and-latency-gains/

33. AI Post Transformers: SGLang: Efficient Language Model Program Execution — Hal Turing & Dr. Ada Shannon, 2025

https://podcast.do-not-panic.com/episodes/sglang-efficient-language-model-program-execution/

34. AI Post Transformers: Speculative Speculative Decoding — Hal Turing & Dr. Ada Shannon, 2026

https://podcast.do-not-panic.com/episodes/2026-03-25-speculative-speculative-decoding-1b7a10.mp3

35. AI Post Transformers: Jet-Nemotron and PostNAS for Faster Long Context — Hal Turing & Dr. Ada Shannon, 2026

https://podcast.do-not-panic.com/episodes/2026-03-24-jet-nemotron-and-postnas-for-faster-long-436381.mp3

36. AI Post Transformers: FlexGen: High-Throughput LLM Inference on a Single GPU — Hal Turing & Dr. Ada Shannon, 2025

https://podcast.do-not-panic.com/episodes/flexgen-high-throughput-llm-inference-on-a-single-gpu/

Interactive Visualization: FlatAttention for Tile-Based Accelerator Inference

This episode explores a 2025 arXiv paper on CXL-based computational memory, focusing on how partial offloading should be structured so applications actually run faster end to end rather than merely showing lower kernel-launch overhead. It explains why the central challenge is coordination between the host CPU and near-memory compute on a remote CXL memory device, especially for memory-bound workloads like graph analytics, sparse retrieval, database-style processing, and KV-heavy inference. The discussion contrasts two existing offloading models—Remote Polling and Bulk Synchronous Flow—arguing that one becomes too chatty while the other introduces lockstep stalls, and that communication semantics such as CXL.io versus CXL.mem fundamentally shape performance. Listeners would find it interesting because it reframes near-memory computing as a systems and scheduling problem, not just a kernel-selection problem, with direct implications for emerging disaggregated memory and CXL deployments.

Sources:

1. Offloading to CXL-based Computational Memory — Suyeon Lee, Kangkyu Park, Kwangsik Shin, Ada Gavrilovska, 2025

http://arxiv.org/abs/2512.04449

2. A Case for Memory-Centric HPC System Design — Dong Li, Jeffrey S. Vetter and others, 2015

https://scholar.google.com/scholar?q=A+Case+for+Memory-Centric+HPC+System+Design

3. The Datacenter as a Computer: Designing Warehouse-Scale Machines, Third Edition — Luiz André Barroso, Urs Hölzle, Parthasarathy Ranganathan, 2018

https://scholar.google.com/scholar?q=The+Datacenter+as+a+Computer:+Designing+Warehouse-Scale+Machines,+Third+Edition

4. A Roofline Model of Energy — Samuel Williams, Andrew Waterman, David Patterson and others, 2014

https://scholar.google.com/scholar?q=A+Roofline+Model+of+Energy

5. M2NDP: A Near-Memory Processing Architecture for CXL Memory Expansion — Authors commonly cited as the M2NDP team; exact author list varies by version, 2024

https://scholar.google.com/scholar?q=M2NDP:+A+Near-Memory+Processing+Architecture+for+CXL+Memory+Expansion

6. M2NDP — Not fully specified in the provided excerpt, Not specified in excerpt

https://scholar.google.com/scholar?q=M2NDP

7. Compute Express Link Specification / CXL 2.0 and 3.0 ecosystem references — Compute Express Link Consortium, 2020-2022

https://scholar.google.com/scholar?q=Compute+Express+Link+Specification+/+CXL+2.0+and+3.0+ecosystem+references

8. AIFM: High-Performance, Application-Integrated Far Memory — Anirudh Suresh et al., 2020

https://scholar.google.com/scholar?q=AIFM:+High-Performance,+Application-Integrated+Far+Memory

9. Infiniswap — Juncheng Gu et al., 2017

https://scholar.google.com/scholar?q=Infiniswap

10. LegoOS: A Disseminated, Distributed OS for Hardware Resource Disaggregation — Yizhou Shan et al., 2019

https://scholar.google.com/scholar?q=LegoOS:+A+Disseminated,+Distributed+OS+for+Hardware+Resource+Disaggregation

11. CXL- and near-memory-processing-related prior offloading works cited as [8], [19], [11], [28], [14], [10], [30], [13], [12], [27], [26], [16] — Various, Various

https://scholar.google.com/scholar?q=CXL-+and+near-memory-processing-related+prior+offloading+works+cited+as+[8],+[19],+[11],+[28],+[14],+[10],+[30],+[13],+[12],+[27],+[26],+[16]

12. Dissecting CXL Memory Performance at Scale: Analysis, Modeling, and Optimization — approx. recent systems authors, exact list unclear from snippet, 2024/2025

https://scholar.google.com/scholar?q=Dissecting+CXL+Memory+Performance+at+Scale:+Analysis,+Modeling,+and+Optimization

13. TRACE: Unlocking Effective CXL Bandwidth via Lossless Compression and Precision Scaling — approx. recent architecture/systems authors, exact list unclear from snippet, 2024/2025

https://scholar.google.com/scholar?q=TRACE:+Unlocking+Effective+CXL+Bandwidth+via+Lossless+Compression+and+Precision+Scaling

14. Remote Memory Prefetching: Is Coarse-grained Fine? — approx. recent systems authors, exact list unclear from snippet, 2024/2025

https://scholar.google.com/scholar?q=Remote+Memory+Prefetching:+Is+Coarse-grained+Fine?

15. IBEX: Internal Bandwidth-Efficient Compression Architecture for Scalable CXL Memory Expansion — approx. recent architecture authors, exact list unclear from snippet, 2024/2025

https://scholar.google.com/scholar?q=IBEX:+Internal+Bandwidth-Efficient+Compression+Architecture+for+Scalable+CXL+Memory+Expansion

16. A Near CXL Memory Processing Architecture for Distributed Graph Neural Network Inference and Training — approx. recent ML-systems/architecture authors, exact list unclear from snippet, 2024/2025

https://scholar.google.com/scholar?q=A+Near+CXL+Memory+Processing+Architecture+for+Distributed+Graph+Neural+Network+Inference+and+Training

17. Scalable Processing-Near-Memory for 1M-Token LLM Inference: CXL-Enabled KV-Cache Management Beyond GPU Limits — approx. recent LLM systems authors, exact list unclear from snippet, 2024/2025

https://scholar.google.com/scholar?q=Scalable+Processing-Near-Memory+for+1M-Token+LLM+Inference:+CXL-Enabled+KV-Cache+Management+Beyond+GPU+Limits

18. Enabling Efficient Large Recommendation Model Training with Near CXL Memory Processing — approx. recent ML systems authors, exact list unclear from snippet, 2024/2025

https://scholar.google.com/scholar?q=Enabling+Efficient+Large+Recommendation+Model+Training+with+Near+CXL+Memory+Processing

19. Towards Continuous Checkpointing for HPC Systems Using CXL — approx. recent HPC/storage systems authors, exact list unclear from snippet, 2024/2025

https://scholar.google.com/scholar?q=Towards+Continuous+Checkpointing+for+HPC+Systems+Using+CXL

20. System Suspend with Asynchronous Resume using CXL-Based Persistent Memory — approx. recent systems authors, exact list unclear from snippet, 2024/2025

https://scholar.google.com/scholar?q=System+Suspend+with+Asynchronous+Resume+using+CXL-Based+Persistent+Memory

21. AI Post Transformers: Xerxes: CXL 3.0 Simulation for Scalable Memory Systems — Hal Turing & Dr. Ada Shannon, 2026

https://podcast.do-not-panic.com/episodes/2026-03-18-xerxes-cxl-30-simulation-for-scalable-me-fdc3f1.mp3

22. AI Post Transformers: CXL-SpecKV: Bridging the LLM Memory Wall with Speculative FPGA Disaggregation — Hal Turing & Dr. Ada Shannon, 2026

https://podcast.do-not-panic.com/episodes/cxl-speckv-bridging-the-llm-memory-wall-with-speculative-fpga-disaggregation/

23. AI Post Transformers: ByteCheckpoint: A Unified LLM Checkpointing System — Hal Turing & Dr. Ada Shannon, 2025

https://podcast.do-not-panic.com/episodes/bytecheckpoint-a-unified-llm-checkpointing-system/

24. AI Post Transformers: Teraio: Cost-Efficient LLM Training via Lifetime-Aware Tensor Offloading — Hal Turing & Dr. Ada Shannon, 2025

https://podcast.do-not-panic.com/episodes/teraio-cost-efficient-llm-training-via-lifetime-aware-tensor-offloading/

Interactive Visualization: CXL Computational Memory Offloading for Lower Runtime

This episode explores a practical systems paper on speeding up Mixture-of-Experts language models at inference time by changing how tokens are routed during decoding, without any retraining. It explains why MoE models, despite using sparse per-token computation, can still be slow in real-world serving because small decode batches activate a large union of different experts, making inference memory-bound due to irregular weight loading. The discussion highlights the paper’s central argument that routing should be batch-aware rather than token-local, so expert choices account for which experts are already being loaded for other tokens in the batch. Listeners would find it interesting for its clear explanation of the gap between MoE’s theoretical efficiency and deployment reality, and for its focus on a low-cost serving optimization with direct economic impact on LLM inference.

Sources:

1. Opportunistic Expert Activation: Batch-Aware Expert Routing for Faster Decode Without Retraining — Costin-Andrei Oncescu, Qingyang Wu, Wai Tong Chung, Robert Wu, Bryan Gopal, Junxiong Wang, Tri Dao, Ben Athiwaratkun, 2025

http://arxiv.org/abs/2511.02237

2. Outrageously Large Neural Networks: The Sparsely-Gated Mixture-of-Experts Layer — Noam Shazeer, Azalia Mirhoseini, Krzysztof Maziarz, Andy Davis, Quoc Le, Geoffrey Hinton, Jeff Dean, 2017

https://scholar.google.com/scholar?q=Outrageously+Large+Neural+Networks:+The+Sparsely-Gated+Mixture-of-Experts+Layer

3. Switch Transformers: Scaling to Trillion Parameter Models with Simple and Efficient Sparsity — William Fedus, Barret Zoph, Noam Shazeer, 2021

https://scholar.google.com/scholar?q=Switch+Transformers:+Scaling+to+Trillion+Parameter+Models+with+Simple+and+Efficient+Sparsity

4. MegaBlocks: Efficient Sparse Training with Mixture-of-Experts — Samyam Rajbhandari, Jeff Rasley, Olatunji Ruwase, Yuxiong He and collaborators, 2023

https://scholar.google.com/scholar?q=MegaBlocks:+Efficient+Sparse+Training+with+Mixture-of-Experts

5. Opportunistic Expert Activation: Batch-Aware Expert Routing for Faster Decode Without Retraining — Costin-Andrei Oncescu, Qingyang Wu, Wai Tong Chung, Robert Wu, Bryan Gopal, Junxiong Wang, Tri Dao, Ben Athiwaratkun, 2025

https://scholar.google.com/scholar?q=Opportunistic+Expert+Activation:+Batch-Aware+Expert+Routing+for+Faster+Decode+Without+Retraining

6. Efficient Memory Management for Large Language Model Serving with PagedAttention — Woosuk Kwon, Zhuohan Li, Siyuan Zhuang, Ying Sheng, Lianmin Zheng, Hao Zhang, Eric Gonzalez, Ion Stoica, Joseph E. Gonzalez, 2023

https://scholar.google.com/scholar?q=Efficient+Memory+Management+for+Large+Language+Model+Serving+with+PagedAttention

7. SGLang: Efficient Execution of Structured Language Model Programs — Lianmin Zheng and collaborators, 2024

https://scholar.google.com/scholar?q=SGLang:+Efficient+Execution+of+Structured+Language+Model+Programs

8. DeepSeek-V3 Technical Report — DeepSeek-AI / Liu et al., 2024

https://scholar.google.com/scholar?q=DeepSeek-V3+Technical+Report

9. Kimi K2 Technical Report — Kimi Team, 2025

https://scholar.google.com/scholar?q=Kimi+K2+Technical+Report

10. Qwen3 Technical Report — Yang et al., 2025

https://scholar.google.com/scholar?q=Qwen3+Technical+Report

11. The Roofline Model: A Pedagogical Tool for Program Analysis and Optimization — Samuel Williams, Andrew Waterman, David Patterson, 2009

https://scholar.google.com/scholar?q=The+Roofline+Model:+A+Pedagogical+Tool+for+Program+Analysis+and+Optimization

12. FlashAttention: Fast and Memory-Efficient Exact Attention with IO-Awareness — Tri Dao, Daniel Y. Fu, Stefano Ermon, Atri Rudra, Christopher Ré, 2022

https://scholar.google.com/scholar?q=FlashAttention:+Fast+and+Memory-Efficient+Exact+Attention+with+IO-Awareness

13. Moe-Infinity: Efficient MoE Inference on Personal Machines with Sparsity-Aware Expert Cache — approx. recent systems paper, exact authors unclear from snippet, 2024/2025

https://scholar.google.com/scholar?q=Moe-Infinity:+Efficient+MoE+Inference+on+Personal+Machines+with+Sparsity-Aware+Expert+Cache

14. Diff-MoE: Efficient Batched MoE Inference with Priority-Driven Differential Expert Caching — approx. recent systems paper, exact authors unclear from snippet, 2024/2025

https://scholar.google.com/scholar?q=Diff-MoE:+Efficient+Batched+MoE+Inference+with+Priority-Driven+Differential+Expert+Caching

15. SliceMoE: Bit-Sliced Expert Caching under Miss-Rate Constraints for Efficient MoE Inference — approx. recent systems paper, exact authors unclear from snippet, 2024/2025

https://scholar.google.com/scholar?q=SliceMoE:+Bit-Sliced+Expert+Caching+under+Miss-Rate+Constraints+for+Efficient+MoE+Inference

16. A Survey on Inference Optimization Techniques for Mixture of Experts Models — approx. recent survey, exact authors unclear from snippet, 2024/2025

https://scholar.google.com/scholar?q=A+Survey+on+Inference+Optimization+Techniques+for+Mixture+of+Experts+Models

17. Rewiring Experts on the Fly: Continuous Rerouting for Better Online Adaptation in Mixture-of-Expert Models — approx. recent paper, exact authors unclear from snippet, 2024/2025

https://scholar.google.com/scholar?q=Rewiring+Experts+on+the+Fly:+Continuous+Rerouting+for+Better+Online+Adaptation+in+Mixture-of-Expert+Models

18. Stabilizing MoE Reinforcement Learning by Aligning Training and Inference Routers — approx. recent paper, exact authors unclear from snippet, 2024/2025

https://scholar.google.com/scholar?q=Stabilizing+MoE+Reinforcement+Learning+by+Aligning+Training+and+Inference+Routers

19. AI Post Transformers: Switch Transformers: Trillion Parameter Models with Sparsity — Hal Turing & Dr. Ada Shannon, 2025

https://podcast.do-not-panic.com/episodes/switch-transformers-trillion-parameter-models-with-sparsity/

20. AI Post Transformers: LFM2-8B-A1B: Efficient On-Device Mixture-of-Experts — Hal Turing & Dr. Ada Shannon, 2025

https://podcast.do-not-panic.com/episodes/lfm2-8b-a1b-efficient-on-device-mixture-of-experts/

21. AI Post Transformers: FlexGen: High-Throughput LLM Inference on a Single GPU — Hal Turing & Dr. Ada Shannon, 2025

https://podcast.do-not-panic.com/episodes/flexgen-high-throughput-llm-inference-on-a-single-gpu/

22. AI Post Transformers: FlashAttention-2: Faster Attention with Better Parallelism — Hal Turing & Dr. Ada Shannon, 2025

https://podcast.do-not-panic.com/episodes/flashattention-2-faster-attention-with-better-parallelism/

23. AI Post Transformers: Lookahead Q-Cache for Consistent KV Eviction — Hal Turing & Dr. Ada Shannon, 2026

https://podcast.do-not-panic.com/episodes/2026-03-25-lookahead-q-cache-for-consistent-kv-evic-d97b09.mp3

24. AI Post Transformers: FAST26: Bidaw: Enhancing Key-Value Caching for Interactive LLM Serving via Bidirectional — Hal Turing & Dr. Ada Shannon, 2026

https://podcast.do-not-panic.com/episodes/fast26-bidaw-enhancing-key-value-caching-for-interactive-llm-serving-via-bidirec/

Interactive Visualization: Batch-Aware Expert Routing for Faster MoE Decoding

This episode explores why AI agents become a fundamentally different security problem once language models can browse the web, read email, call tools, store memory, and act inside real software environments. It explains prompt injection as the core boundary failure, showing how webpages, emails, retrieved notes, or API responses can be mistaken for trusted instructions, turning ordinary content into an attack vector with real operational consequences. The discussion then sharpens the distinction between one-off prompt attacks and more systemic failures such as memory poisoning and multi-agent compromise, where corrupted state can persist across sessions or spread through delegated workflows. A listener would find it interesting because it frames agent safety as a concrete systems-security challenge, not just a model-behavior quirk, and clarifies why greater capability also widens the blast radius of failure.

Sources:

1. AI Agent Traps and Prompt Injection

/tmp/submission-source-_s144w4z.txt

2. Ignore Previous Prompt: Attack Techniques For Language Models — Fábio Perez, Ian Ribeiro, 2022

https://scholar.google.com/scholar?q=Ignore+Previous+Prompt:+Attack+Techniques+For+Language+Models

3. Not what you've signed up for: Compromising Real-World LLM-Integrated Applications with Indirect Prompt Injection — Kai Greshake, Sahar Abdelnabi, Shailesh Mishra, Christoph Endres, Thorsten Holz, Mario Fritz, 2023

https://scholar.google.com/scholar?q=Not+what+you've+signed+up+for:+Compromising+Real-World+LLM-Integrated+Applications+with+Indirect+Prompt+Injection

4. Prompt Injection attack against LLM-integrated Applications — Yi Liu, Gelei Deng, Yuekang Li, Kailong Wang, Tianwei Zhang, Yepang Liu, Haoyu Wang, Yan Zheng, Yang Liu, 2023

https://scholar.google.com/scholar?q=Prompt+Injection+attack+against+LLM-integrated+Applications

5. Prompt Injection Attacks and Defenses in LLM-Integrated Applications — Yupei Liu, Yuqi Jia, Runpeng Geng, Jinyuan Jia, Neil Zhenqiang Gong, 2023

https://scholar.google.com/scholar?q=Prompt+Injection+Attacks+and+Defenses+in+LLM-Integrated+Applications

6. AI Agents Under Threat: A Survey of Key Security Challenges and Future Pathways — Zehang Deng, Yongjian Guo, Changzhou Han, Wanlun Ma, Junwu Xiong, Sheng Wen, Yang Xiang, 2024

https://scholar.google.com/scholar?q=AI+Agents+Under+Threat:+A+Survey+of+Key+Security+Challenges+and+Future+Pathways

7. Open Challenges in Multi-Agent Security: Towards Secure Systems of Interacting AI Agents — Christian Schroeder de Witt, 2025

https://scholar.google.com/scholar?q=Open+Challenges+in+Multi-Agent+Security:+Towards+Secure+Systems+of+Interacting+AI+Agents

8. Red-Teaming LLM Multi-Agent Systems via Communication Attacks — Pengfei He, Yupin Lin, Shen Dong, Han Xu, Yue Xing, Hui Liu, 2025

https://scholar.google.com/scholar?q=Red-Teaming+LLM+Multi-Agent+Systems+via+Communication+Attacks

9. G-Safeguard: A Topology-Guided Security Lens and Treatment on LLM-based Multi-agent Systems — Shilong Wang, Guibin Zhang, Miao Yu, Guancheng Wan, Fanci Meng, Chongye Guo, Kun Wang, Yang Wang, 2025

https://scholar.google.com/scholar?q=G-Safeguard:+A+Topology-Guided+Security+Lens+and+Treatment+on+LLM-based+Multi-agent+Systems

10. InjecAgent: Benchmarking Indirect Prompt Injections in Tool-Integrated Large Language Model Agents — Qiusi Zhan, Zhixiang Liang, Zifan Ying, Daniel Kang, 2024

https://scholar.google.com/scholar?q=InjecAgent:+Benchmarking+Indirect+Prompt+Injections+in+Tool-Integrated+Large+Language+Model+Agents

11. Agent Security Bench (ASB): Formalizing and Benchmarking Attacks and Defenses in LLM-based Agents — Hanrong Zhang, Jingyuan Huang, Kai Mei, Yifei Yao, Zhenting Wang, Chenlu Zhan, Hongwei Wang, Yongfeng Zhang, 2024

https://scholar.google.com/scholar?q=Agent+Security+Bench+(ASB):+Formalizing+and+Benchmarking+Attacks+and+Defenses+in+LLM-based+Agents

12. AgentPoison: Red-teaming LLM Agents via Poisoning Memory or Knowledge Bases — Zhaorun Chen, Zhen Xiang, Chaowei Xiao, Dawn Song, Bo Li, 2024

https://scholar.google.com/scholar?q=AgentPoison:+Red-teaming+LLM+Agents+via+Poisoning+Memory+or+Knowledge+Bases

13. Prompt Infection: LLM-to-LLM Prompt Injection within Multi-Agent Systems — Donghyun Lee, Mo Tiwari, 2024

https://scholar.google.com/scholar?q=Prompt+Infection:+LLM-to-LLM+Prompt+Injection+within+Multi-Agent+Systems

14. Multi-Agent Systems Execute Arbitrary Malicious Code — Harold Triedman, Rishi D. Jha, Vitaly Shmatikov, 2025

https://scholar.google.com/scholar?q=Multi-Agent+Systems+Execute+Arbitrary+Malicious+Code

15. Prompt Injection Attacks on Large Language Models: A Survey of Attack Methods, Root Causes, and Defense Strategies — approx. survey by prompt-injection/security researchers, 2025

https://scholar.google.com/scholar?q=Prompt+Injection+Attacks+on+Large+Language+Models:+A+Survey+of+Attack+Methods,+Root+Causes,+and+Defense+Strategies

16. Prompting for LLM Security and RAG: A Survey from Zero-Shot to Automatic Prompt Optimization (APO) and Prompt-Injection Defenses — approx. security/RAG survey authors, 2025

https://scholar.google.com/scholar?q=Prompting+for+LLM+Security+and+RAG:+A+Survey+from+Zero-Shot+to+Automatic+Prompt+Optimization+(APO)+and+Prompt-Injection+Defenses

17. Veriguard: Enhancing LLM Agent Safety via Verified Code Generation — approx. systems/security authors, 2025

https://scholar.google.com/scholar?q=Veriguard:+Enhancing+LLM+Agent+Safety+via+Verified+Code+Generation

18. Enforcement Agents: Enhancing Accountability and Resilience in Multi-Agent AI Frameworks — approx. multi-agent safety authors, 2025

https://scholar.google.com/scholar?q=Enforcement+Agents:+Enhancing+Accountability+and+Resilience+in+Multi-Agent+AI+Frameworks

19. Monitoring LLM-Based Multi-Agent Systems Against Corruptions via Node Evaluation — approx. multi-agent monitoring authors, 2025

https://scholar.google.com/scholar?q=Monitoring+LLM-Based+Multi-Agent+Systems+Against+Corruptions+via+Node+Evaluation

20. Enhancing Robustness of LLM-Driven Multi-Agent Systems Through Randomized Smoothing — approx. robustness/safety authors, 2025

https://scholar.google.com/scholar?q=Enhancing+Robustness+of+LLM-Driven+Multi-Agent+Systems+Through+Randomized+Smoothing

21. Assessing and Enhancing the Robustness of LLM-Based Multi-Agent Systems Through Chaos Engineering — approx. systems robustness authors, 2025

https://scholar.google.com/scholar?q=Assessing+and+Enhancing+the+Robustness+of+LLM-Based+Multi-Agent+Systems+Through+Chaos+Engineering

22. AI Post Transformers: Memory in the Age of AI Agents: Forms, Functions, Dynamics — Hal Turing & Dr. Ada Shannon, 2026

https://podcast.do-not-panic.com/episodes/2026-03-16-memory-in-the-age-of-ai-agents-forms-fun-5abc60.mp3

23. AI Post Transformers: NeurIPS 2025: Agentic Plan Caching: Test-Time Memory for Fast and Cost-Efficient LLM Agents — Hal Turing & Dr. Ada Shannon, 2026

https://podcast.do-not-panic.com/episodes/neurips-2025-agentic-plan-caching-test-time-memory-for-fast-and-cost-efficient-l/

24. AI Post Transformers: ReasoningBank: Scaling Agent Self-Evolving with Reasoning Memory — Hal Turing & Dr. Ada Shannon, 2025

https://podcast.do-not-panic.com/episodes/reasoningbank-scaling-agent-self-evolving-with-reasoning-memory/

25. AI Post Transformers: Qwen3Guard: Streaming Three-Way Safety Classification for LLMs — Hal Turing & Dr. Ada Shannon, 2026

https://podcast.do-not-panic.com/episodes/2026-03-16-qwen3guard-streaming-three-way-safety-cl-26b0ef.mp3

Interactive Visualization: AI Agent Traps and Prompt Injection

This episode explores a paper on how generative multi-agent systems can develop failure modes that do not appear when models are evaluated one at a time. It explains how planner-worker-reviewer loops, negotiation setups, handoff chains, and committee-style aggregation can produce system-level problems such as strategic manipulation, collusion-like behavior, misreporting, conformity, and biased group decisions. The discussion focuses on the paper’s three main risk families: incentive exploitation, collective-cognition failures, and governance breakdowns, while also unpacking the benchmark scenarios used to test those dynamics. Listeners would find it interesting because it connects current real-world agent orchestration patterns to concrete safety and reliability risks, while also probing whether the paper’s evidence is strong enough in light of limited statistics and missing baseline comparisons.

Sources:

1. Emergent Social Intelligence Risks in Generative Multi-Agent Systems — Yue Huang, Yu Jiang, Wenjie Wang, Haomin Zhuang, Xiaonan Luo, Yuchen Ma, Zhangchen Xu, Zichen Chen, Nuno Moniz, Zinan Lin, Pin-Yu Chen, Nitesh V Chawla, Nouha Dziri, Huan Sun, Xiangliang Zhang, 2026

http://arxiv.org/abs/2603.27771

2. CAMEL: Communicative Agents for "Mind" Exploration of Large Language Model Society — Guohao Li, Hasan Abed Al Kader Hammoud, Hani Itani, Dmitrii Khizbullin, Bernard Ghanem, 2023

https://scholar.google.com/scholar?q=CAMEL:+Communicative+Agents+for+"Mind"+Exploration+of+Large+Language+Model+Society

3. AutoGen: Enabling Next-Gen LLM Applications via Multi-Agent Conversation — Qingyun Wu, Gagan Bansal, Jieyu Zhang, Yiran Wu, Beibin Li, Erkang Zhu, Li Jiang, Xiaoyun Zhang, Shaokun Zhang, Ahmed Awadallah, Ryen W. White, Doug Burger, Chi Wang, 2024

https://scholar.google.com/scholar?q=AutoGen:+Enabling+Next-Gen+LLM+Applications+via+Multi-Agent+Conversation

4. MetaGPT: Meta Programming for A Multi-Agent Collaborative Framework — Sirui Hong, Mingchen Zhuge, Jiaqi Chen, Xiawu Zheng, Yuheng Cheng, Ceyao Zhang, Jinlin Wang, Zili Wang, Steven Ka Shing Yau, Zijuan Lin, Liyang Zhou, Chenyu Ran, Lingfeng Xiao, Chenglin Wu, Jürgen Schmidhuber, 2023

https://scholar.google.com/scholar?q=MetaGPT:+Meta+Programming+for+A+Multi-Agent+Collaborative+Framework

5. Large Language Model based Multi-Agents: A Survey of Progress and Challenges — Taicheng Guo, Xiuying Chen, Yaqi Wang, Ruidi Chang, Shichao Pei, Nitesh V. Chawla, Olaf Wiest, Xiangliang Zhang, 2024

https://scholar.google.com/scholar?q=Large+Language+Model+based+Multi-Agents:+A+Survey+of+Progress+and+Challenges

6. Generative Agents: Interactive Simulacra of Human Behavior — Joon Sung Park, Joseph C. O'Brien, Carrie J. Cai, Meredith Ringel Morris, Percy Liang, Michael S. Bernstein, 2023

https://scholar.google.com/scholar?q=Generative+Agents:+Interactive+Simulacra+of+Human+Behavior

7. AgentVerse: Facilitating Multi-Agent Collaboration and Exploring Emergent Behaviors — Weize Chen, Yusheng Su, Jingwei Zuo, Cheng Yang, Chenfei Yuan, Chi-Min Chan, Heyang Yu, Yaxi Lu, Yi-Hsin Hung, Chen Qian, Yujia Qin, Xin Cong, Ruobing Xie, Zhiyuan Liu, Maosong Sun, Jie Zhou, 2023

https://scholar.google.com/scholar?q=AgentVerse:+Facilitating+Multi-Agent+Collaboration+and+Exploring+Emergent+Behaviors

8. Persona Inconstancy in Multi-Agent LLM Collaboration: Conformity, Confabulation, and Impersonation — Razan Baltaji, Babak Hemmatian, Lav R. Varshney, 2024

https://scholar.google.com/scholar?q=Persona+Inconstancy+in+Multi-Agent+LLM+Collaboration:+Conformity,+Confabulation,+and+Impersonation

9. Multi-Agent Risks from Advanced AI — Lewis Hammond, Alan Chan, Jesse Clifton, Jason Hoelscher-Obermaier and many coauthors, 2025

https://scholar.google.com/scholar?q=Multi-Agent+Risks+from+Advanced+AI

10. Autonomous Algorithmic Collusion: Q-Learning Under Sequential Pricing — Timo Klein, 2019

https://scholar.google.com/scholar?q=Autonomous+Algorithmic+Collusion:+Q-Learning+Under+Sequential+Pricing

11. Artificial Intelligence, Algorithmic Pricing, and Collusion — Emilio Calvano, Giacomo Calzolari, Vincenzo Denicolò, Sergio Pastorello, 2020

https://scholar.google.com/scholar?q=Artificial+Intelligence,+Algorithmic+Pricing,+and+Collusion

12. Strategic Collusion of LLM Agents: Market Division in Multi-Commodity Competitions — Ryan Y. Lin, Siddhartha Ojha, Kevin Cai, Maxwell F. Chen, 2024

https://scholar.google.com/scholar?q=Strategic+Collusion+of+LLM+Agents:+Market+Division+in+Multi-Commodity+Competitions

13. AI-Powered Trading, Algorithmic Collusion, and Price Efficiency — Winston Wei Dou, Itay Goldstein, Yan Ji, 2025

https://scholar.google.com/scholar?q=AI-Powered+Trading,+Algorithmic+Collusion,+and+Price+Efficiency

14. Emergence of Social Norms in Generative Agent Societies: Principles and Architecture — Siyue Ren, Zhiyao Cui, Ruiqi Song, Zhen Wang, Shuyue Hu, 2024

https://scholar.google.com/scholar?q=Emergence+of+Social+Norms+in+Generative+Agent+Societies:+Principles+and+Architecture

15. Algorithmic Collusion at Test Time: A Meta-game Design and Evaluation — Yuhong Luo, Daniel Schoepflin, Xintong Wang, 2026

https://scholar.google.com/scholar?q=Algorithmic+Collusion+at+Test+Time:+A+Meta-game+Design+and+Evaluation

16. NetSafe: Exploring the Topological Safety of Multi-agent System — Miao Yu et al., 2025

https://scholar.google.com/scholar?q=NetSafe:+Exploring+the+Topological+Safety+of+Multi-agent+System

17. Institutional AI: Governing LLM Collusion in Multi-Agent Cournot Markets via Public Governance Graphs — Marcantonio Bracale Syrnikov et al., 2026

https://scholar.google.com/scholar?q=Institutional+AI:+Governing+LLM+Collusion+in+Multi-Agent+Cournot+Markets+via+Public+Governance+Graphs

18. Verification-Aware Planning for Multi-Agent Systems — Tianyang Xu, Dan Zhang, Kushan Mitra, Estevam Hruschka, 2025

https://scholar.google.com/scholar?q=Verification-Aware+Planning+for+Multi-Agent+Systems

19. State and Memory is All You Need for Robust and Reliable AI Agents — Matthew Muhoberac et al., 2025

https://scholar.google.com/scholar?q=State+and+Memory+is+All+You+Need+for+Robust+and+Reliable+AI+Agents

20. AI Post Transformers: Multiagent Debate Improves Language Model Reasoning — Hal Turing & Dr. Ada Shannon, 2025

https://podcast.do-not-panic.com/episodes/multiagent-debate-improves-language-model-reasoning/

21. AI Post Transformers: Memory in the Age of AI Agents: Forms, Functions, Dynamics — Hal Turing & Dr. Ada Shannon, 2026

https://podcast.do-not-panic.com/episodes/2026-03-16-memory-in-the-age-of-ai-agents-forms-fun-5abc60.mp3

22. AI Post Transformers: Qwen3Guard: Streaming Three-Way Safety Classification for LLMs — Hal Turing & Dr. Ada Shannon, 2026

https://podcast.do-not-panic.com/episodes/2026-03-16-qwen3guard-streaming-three-way-safety-cl-26b0ef.mp3

23. AI Post Transformers: Tree-based Group Policy Optimization for LLM Agents — Hal Turing & Dr. Ada Shannon, 2025

https://podcast.do-not-panic.com/episodes/tree-based-group-policy-optimization-for-llm-agents/

24. AI Post Transformers: Mem0: Scalable Long-Term Memory for AI Agents — Hal Turing & Dr. Ada Shannon, 2025

https://podcast.do-not-panic.com/episodes/mem0-scalable-long-term-memory-for-ai-agents/

This episode explores Meta-Harness, a paper arguing that a large share of LLM system performance comes from the surrounding harness code that manages memory, retrieval, tool use, context formatting, and control flow rather than from model weights alone. It explains how the method uses an outer-loop coding agent to rewrite harness code, inspect raw traces and logs stored on disk, and search for better system designs across tasks like text classification, retrieval-based math reasoning, and agentic coding. The discussion highlights why this matters: in multi-step systems, the same fixed model can perform very differently depending on what information it sees, when it sees it, and how the wrapper code structures the interaction. Listeners would find it interesting because it reframes progress in AI systems as a systems-engineering problem, raising the possibility that better scaffolding around existing models may unlock major gains without retraining the models themselves.

Sources:

1. Meta-Harness: End-to-End Optimization of Model Harnesses — Yoonho Lee, Roshen Nair, Qizheng Zhang, Kangwook Lee, Omar Khattab, Chelsea Finn, 2026

http://arxiv.org/abs/2603.28052

2. https://yoonholee.com/meta-harness/

https://yoonholee.com/meta-harness/

3. DSPy: Compiling Declarative Language Model Calls into Self-Improving Pipelines — Omar Khattab, Arnav Singhvi, Paridhi Maheshwari, Zhiyuan Zhang, Keshav Santhanam, Matei Zaharia, Christopher Potts, and others, 2023

https://scholar.google.com/scholar?q=DSPy:+Compiling+Declarative+Language+Model+Calls+into+Self-Improving+Pipelines

4. Reflexion: Language Agents with Verbal Reinforcement Learning — Noah Shinn, Federico Cassano, Edward Berman, Ashwin Gopinath, Karthik Narasimhan, Shunyu Yao, 2023

https://scholar.google.com/scholar?q=Reflexion:+Language+Agents+with+Verbal+Reinforcement+Learning

5. MemGPT: Towards LLMs as Operating Systems — Charles Packer, Sarah Wooders, Kevin Lin, Vivian Fang, Shishir G. Patil, Ion Stoica, Joseph E. Gonzalez, 2023

https://scholar.google.com/scholar?q=MemGPT:+Towards+LLMs+as+Operating+Systems

6. Agentic Context Engineering: Evolving Contexts for Self-Improving Language Models — Qizheng Zhang, Changran Hu, Shubhangi Upasani, Boyuan Ma, Fenglu Hong, James Zou, Kunle Olukotun, and others, 2025

https://scholar.google.com/scholar?q=Agentic+Context+Engineering:+Evolving+Contexts+for+Self-Improving+Language+Models

7. GEPA: Reflective Prompt Evolution Can Outperform Reinforcement Learning — Lakshya A. Agrawal, Shangyin Tan, Dilara Soylu, Noah Ziems, Rishi Khare, Krista Opsahl-Ong, Arnav Singhvi, Herumb Shandilya, Michael J. Ryan, Meng Jiang, Christopher Potts, Koushik Sen, Alexandros G. Dimakis, Ion Stoica, Dan Klein, Matei Zaharia, Omar Khattab, 2025

https://scholar.google.com/scholar?q=GEPA:+Reflective+Prompt+Evolution+Can+Outperform+Reinforcement+Learning

8. TextGrad: Automatic "Differentiation" via Text — Mert Yuksekgonul, Federico Bianchi, Joseph Boen, Sheng Liu, Zhi Huang, Carlos Guestrin, James Zou, 2024

https://scholar.google.com/scholar?q=TextGrad:+Automatic+"Differentiation"+via+Text

9. Large Language Models as Optimizers — Chengrun Yang, Xuezhi Wang, Yifeng Lu, Hanxiao Liu, Quoc V. Le, Denny Zhou, Xinyun Chen, 2023

https://scholar.google.com/scholar?q=Large+Language+Models+as+Optimizers

10. AlphaEvolve: A Coding Agent for Scientific and Algorithmic Discovery — Alexander Novikov, Ngan Vu, Marvin Eisenberger, Emilien Dupont, Po-Sen Huang, Adam Zsolt Wagner, Sergey Shirobokov, Borislav Kozlovskii, Francisco J. R. Ruiz, Abbas Mehrabian, M. Pawan Kumar, Abigail See, Swarat Chaudhuri, George Holland, Alex Davies, Sebastian Nowozin, Pushmeet Kohli, Matej Balog, 2025

https://scholar.google.com/scholar?q=AlphaEvolve:+A+Coding+Agent+for+Scientific+and+Algorithmic+Discovery

11. Learning to Discover at Test Time — Mert Yuksekgonul, Daniel Koceja, Xinhao Li, Federico Bianchi, Jed McCaleb, Xiaolong Wang, Jan Kautz, Yejin Choi, James Zou, Carlos Guestrin, Yu Sun, 2026

https://scholar.google.com/scholar?q=Learning+to+Discover+at+Test+Time

12. Grounded Test-Time Adaptation for LLM Agents — Arthur Chen et al., 2025

https://scholar.google.com/scholar?q=Grounded+Test-Time+Adaptation+for+LLM+Agents

13. Evo-Memory: Benchmarking LLM Agent Test-time Learning with Self-Evolving Memory — Tianxin Wei et al., 2025

https://scholar.google.com/scholar?q=Evo-Memory:+Benchmarking+LLM+Agent+Test-time+Learning+with+Self-Evolving+Memory

14. M^2: Dual-Memory Augmentation for Long-Horizon Web Agents via Trajectory Summarization and Insight Retrieval — Dawei Yan et al., 2026

https://scholar.google.com/scholar?q=M^2:+Dual-Memory+Augmentation+for+Long-Horizon+Web+Agents+via+Trajectory+Summarization+and+Insight+Retrieval

15. Reinforcement Fine-Tuning for History-Aware Dense Retriever in RAG — Yicheng Zhang et al., 2026

https://scholar.google.com/scholar?q=Reinforcement+Fine-Tuning+for+History-Aware+Dense+Retriever+in+RAG

16. MAIN-RAG: Multi-Agent Filtering Retrieval-Augmented Generation — Chia-Yuan Chang et al., 2024

https://scholar.google.com/scholar?q=MAIN-RAG:+Multi-Agent+Filtering+Retrieval-Augmented+Generation

17. Fine-tuning with RAG for Improving LLM Learning of New Skills — Humaid Ibrahim, Nikolai Rozanov, Marek Rei, 2025

https://scholar.google.com/scholar?q=Fine-tuning+with+RAG+for+Improving+LLM+Learning+of+New+Skills

18. AI Post Transformers: Agentic Context Engineering: Evolving Contexts for Self-Improving Language Models — Hal Turing & Dr. Ada Shannon, 2026

https://podcast.do-not-panic.com/episodes/agentic-context-engineering-evolving-contexts-for-self-improving-language-models/

19. AI Post Transformers: Mem0: Scalable Long-Term Memory for AI Agents — Hal Turing & Dr. Ada Shannon, 2025

https://podcast.do-not-panic.com/episodes/mem0-scalable-long-term-memory-for-ai-agents/

20. AI Post Transformers: Agentic AI and the Next Intelligence Explosion — Hal Turing & Dr. Ada Shannon, 2026

https://podcast.do-not-panic.com/episodes/2026-03-28-agentic-ai-and-the-next-intelligence-exp-d06561.mp3

This episode explores a March 19, 2026 study on whether large language models respond to out-of-distribution prompts by compressing their internal activity into fewer active dimensions. It explains how the paper connects two traditions in AI research, mechanistic interpretability and representation geometry, by proposing hidden-state sparsity as a measurable internal signature of stress from harder reasoning tasks, longer contexts, and conflicting information. The discussion breaks down the paper’s core metrics, including Top-k Energy and L1 norm, and clarifies why sparser activations should not be treated as proof of better reasoning or cleaner representations. Listeners would find it interesting because it ties abstract internal model behavior to practical questions about robustness, reliability, and how to evaluate language models beyond just whether their final answers look correct.

Sources:

1. Farther the Shift, Sparser the Representation: Analyzing OOD Mechanisms in LLMs — Mingyu Jin, Yutong Yin, Jingcheng Niu, Qingcheng Zeng, Wujiang Xu, Mengnan Du, Wei Cheng, Zhaoran Wang, Tianlong Chen, Dimitris N. Metaxas, 2026

http://arxiv.org/abs/2603.03415

2. Domain Generalization: A Survey — Kaiyang Zhou, Ziwei Liu, Yu Qiao, Tao Xiang, Chen Change Loy, 2021

https://scholar.google.com/scholar?q=Domain+Generalization:+A+Survey

3. Invariant Risk Minimization — Martin Arjovsky, Leon Bottou, Ishaan Gulrajani, David Lopez-Paz, 2019

https://scholar.google.com/scholar?q=Invariant+Risk+Minimization

4. In Search of Lost Domain Generalization — Ishaan Gulrajani, David Lopez-Paz, 2021

https://scholar.google.com/scholar?q=In+Search+of+Lost+Domain+Generalization

5. WILDS: A Benchmark of in-the-Wild Distribution Shifts — Pang Wei Koh, Shiori Sagawa, Henrik Marklund and many others, 2021

https://scholar.google.com/scholar?q=WILDS:+A+Benchmark+of+in-the-Wild+Distribution+Shifts

6. Emergence of Simple-Cell Receptive Field Properties by Learning a Sparse Code for Natural Images — Bruno A. Olshausen, David J. Field, 1996

https://scholar.google.com/scholar?q=Emergence+of+Simple-Cell+Receptive+Field+Properties+by+Learning+a+Sparse+Code+for+Natural+Images

7. Deep Sparse Rectifier Neural Networks — Xavier Glorot, Antoine Bordes, Yoshua Bengio, 2011

https://scholar.google.com/scholar?q=Deep+Sparse+Rectifier+Neural+Networks

8. Intrinsic Dimensionality Explains the Effectiveness of Language Model Fine-Tuning — Armen Aghajanyan, Sonal Gupta, Luke Zettlemoyer, 2021

https://scholar.google.com/scholar?q=Intrinsic+Dimensionality+Explains+the+Effectiveness+of+Language+Model+Fine-Tuning

9. Towards Monosemanticity: Decomposing Language Models With Dictionary Learning — Trenton Bricken, Adly Templeton, Joshua Batson and many others, 2023

https://scholar.google.com/scholar?q=Towards+Monosemanticity:+Decomposing+Language+Models+With+Dictionary+Learning

10. Understanding Intermediate Layers Using Linear Classifier Probes — Guillaume Alain, Yoshua Bengio, 2017

https://scholar.google.com/scholar?q=Understanding+Intermediate+Layers+Using+Linear+Classifier+Probes

11. Deep Contextualized Word Representations — Matthew E. Peters, Mark Neumann, Mohit Iyyer and others, 2018

https://scholar.google.com/scholar?q=Deep+Contextualized+Word+Representations

12. A Structural Probe for Finding Syntax in Word Representations — John Hewitt, Christopher D. Manning, 2019

https://scholar.google.com/scholar?q=A+Structural+Probe+for+Finding+Syntax+in+Word+Representations

13. How Contextual Are Contextualized Word Representations? Comparing the Geometry of BERT, ELMo, and GPT-2 Embeddings — Kawin Ethayarajh, 2019

https://scholar.google.com/scholar?q=How+Contextual+Are+Contextualized+Word+Representations?+Comparing+the+Geometry+of+BERT,+ELMo,+and+GPT-2+Embeddings

14. The Geometry of Innocent Flesh on the Bone: Syntactic Structure in Sentence Embeddings — John Hewitt and Christopher D. Manning, 2019

https://scholar.google.com/scholar?q=The+Geometry+of+Innocent+Flesh+on+the+Bone:+Syntactic+Structure+in+Sentence+Embeddings

15. What Factors Affect the Success of In-Context Learning? Investigating the Role of Model Architecture and Task Features — Jason Wei, Yi Tay, Quoc V. Le, Denny Zhou and others, 2022

https://scholar.google.com/scholar?q=What+Factors+Affect+the+Success+of+In-Context+Learning?+Investigating+the+Role+of+Model+Architecture+and+Task+Features

16. Let's Verify Step by Step — Hunter Lightman, Vineet Kosaraju, Yura Burda, Harri Edwards, Bowen Baker and others, 2024

https://scholar.google.com/scholar?q=Let's+Verify+Step+by+Step

17. Do Language Models Generalize to Longer Contexts? — Yixiao Li and collaborators, 2025

https://scholar.google.com/scholar?q=Do+Language+Models+Generalize+to+Longer+Contexts?

18. Parameter-Efficient Prompt Tuning Makes Generalized and Calibrated Language Models — Nicola De Cao, Wilker Aziz and Ivan Titov, 2022

https://scholar.google.com/scholar?q=Parameter-Efficient+Prompt+Tuning+Makes+Generalized+and+Calibrated+Language+Models

19. The Lottery Ticket Hypothesis: Finding Sparse, Trainable Neural Networks — Jonathan Frankle and Michael Carbin, 2019

https://scholar.google.com/scholar?q=The+Lottery+Ticket+Hypothesis:+Finding+Sparse,+Trainable+Neural+Networks

20. Adaptive Mixtures of Local Experts — Robert A. Jacobs, Michael I. Jordan, Steven J. Nowlan and Geoffrey E. Hinton, 1991

https://scholar.google.com/scholar?q=Adaptive+Mixtures+of+Local+Experts

21. Curriculum Demonstration Selection for In-Context Learning — approx. recent ICL curriculum-learning authors, recent

https://scholar.google.com/scholar?q=Curriculum+Demonstration+Selection+for+In-Context+Learning

22. Let's Learn Step by Step: Enhancing In-Context Learning Ability with Curriculum Learning — approx. recent ICL curriculum-learning authors, recent

https://scholar.google.com/scholar?q=Let's+Learn+Step+by+Step:+Enhancing+In-Context+Learning+Ability+with+Curriculum+Learning

23. Sparse but not Simpler: A Multi-Level Interpretability Analysis of Vision Transformers — approx. recent interpretability authors, recent

https://scholar.google.com/scholar?q=Sparse+but+not+Simpler:+A+Multi-Level+Interpretability+Analysis+of+Vision+Transformers

24. Weight-Sparse Transformers Have Interpretable Circuits — approx. recent mechanistic interpretability authors, recent

https://scholar.google.com/scholar?q=Weight-Sparse+Transformers+Have+Interpretable+Circuits

25. AI Post Transformers: Chain-of-Thought Reasoning: A Brittle Mirage? — Hal Turing & Dr. Ada Shannon, 2025

https://podcast.do-not-panic.com/episodes/chain-of-thought-reasoning-a-brittle-mirage/

26. AI Post Transformers: Advancing Mechanistic Interpretability with Sparse Autoencoders — Hal Turing & Dr. Ada Shannon, 2026

https://podcast.do-not-panic.com/episodes/advancing-mechanistic-interpretability-with-sparse-autoencoders/

27. AI Post Transformers: Measuring LLM Reasoning Effort via Deep-Thinking Tokens — Hal Turing & Dr. Ada Shannon, 2026

https://podcast.do-not-panic.com/episodes/measuring-llm-reasoning-effort-via-deep-thinking-tokens/

28. AI Post Transformers: CLUE: Hidden-State Clustering for Non-parametric Verification — Hal Turing & Dr. Ada Shannon, 2025

https://podcast.do-not-panic.com/episodes/clue-hidden-state-clustering-for-non-parametric-verification/

29. AI Post Transformers: Inverse IFEval: Unlearning LLM Cognitive Inertia — Hal Turing & Dr. Ada Shannon, 2025

https://podcast.do-not-panic.com/episodes/inverse-ifeval-unlearning-llm-cognitive-inertia/

30. AI Post Transformers: Hyper-Scaling LLM Inference with KV Cache Compression — Hal Turing & Dr. Ada Shannon, 2025

https://podcast.do-not-panic.com/episodes/hyper-scaling-llm-inference-with-kv-cache-compression/