# \#papers **URL:** https://www.artofsm.art/tag/papers/315.md [Latest](https://www.artofsm.art/latest.md) · [Categories](https://www.artofsm.art/categories.md) · [Tags](https://www.artofsm.art/tags.md) --- ## [This AI Agent Builds Its OWN Tools to Finish Tasks! (INSANE Agent S4 Test)](https://www.artofsm.art/t/this-ai-agent-builds-its-own-tools-to-finish-tasks-insane-agent-s4-test/26744)
**Author:** [@merefield](https://www.artofsm.art/u/merefield)\ **Replies:** 1\ **Last updated:** [23 October 2026 18:50 UTC](https://www.artofsm.art/t/this-ai-agent-builds-its-own-tools-to-finish-tasks-insane-agent-s4-test/26744 "2026-10-23T18:50:38Z")
The video explores Agent S4, an AI system that improves task efficiency by autonomously creating and refining its own tools, enabling better generalization and faster completion compared to traditional fixe… --- ## [Read a Research Paper in 2 minutes with FREE AI tools](https://www.artofsm.art/t/read-a-research-paper-in-2-minutes-with-free-ai-tools/26424)
**Author:** [@artesia](https://www.artofsm.art/u/artesia)\ **Replies:** 1\ **Last updated:** [20 October 2026 01:54 UTC](https://www.artofsm.art/t/read-a-research-paper-in-2-minutes-with-free-ai-tools/26424 "2026-10-20T01:54:20Z")
The video highlights the challenges of thoroughly reading research papers and introduces several AI tools—such as Google Scholar’s Quick Read, Semantic Scholar’s Semantic Reader, Google Notebook, and the Go… --- ## [AI Now Builds Its Own Training Worlds. I Rebuilt One](https://www.artofsm.art/t/ai-now-builds-its-own-training-worlds-i-rebuilt-one/26107)
**Author:** [@merefield](https://www.artofsm.art/u/merefield)\ **Replies:** 1\ **Last updated:** [17 October 2026 07:54 UTC](https://www.artofsm.art/t/ai-now-builds-its-own-training-worlds-i-rebuilt-one/26107 "2026-10-17T07:54:35Z")
The video reviews the Compo World paper, which introduces a system for training AI agents using a library of reusable services linked by a dependency graph to create complex tasks, but highlights challenges… --- ## [AI News for Academics: Google Deep Review, 257 AI Papers Pulled & an AI Idea Detector](https://www.artofsm.art/t/ai-news-for-academics-google-deep-review-257-ai-papers-pulled-an-ai-idea-detector/26023)
**Author:** [@artesia](https://www.artofsm.art/u/artesia)\ **Replies:** 1\ **Last updated:** [16 October 2026 02:41 UTC](https://www.artofsm.art/t/ai-news-for-academics-google-deep-review-257-ai-papers-pulled-an-ai-idea-detector/26023 "2026-10-16T02:41:57Z")
This week’s AI news for academia covers controversies like Nicholas Pollson’s record 257 AI-generated papers, many removed for quality concerns, alongside the development of tools like Idea Lens to detect A… --- ## [OpenAI's AI Model Publishes Hundreds of Mathematical Manuscripts, Sparking Debate and Breakthroughs](https://www.artofsm.art/t/openais-ai-model-publishes-hundreds-of-mathematical-manuscripts-sparking-debate-and-breakthroughs/26015)
**Author:** [@artesia](https://www.artofsm.art/u/artesia)\ **Replies:** 0\ **Last updated:** [15 October 2026 21:51 UTC](https://www.artofsm.art/t/openais-ai-model-publishes-hundreds-of-mathematical-manuscripts-sparking-debate-and-breakthroughs/26015 "2026-10-15T21:51:43Z")
OpenAI has released over 700 mathematical manuscripts generated by its advanced AI model, marking a significant milestone in the intersection of artificial intelligence and mathematics. These manuscripts, which includ… --- ## [AI is solving math faster than human verification](https://www.artofsm.art/t/ai-is-solving-math-faster-than-human-verification/25862)
**Author:** [@artesia](https://www.artofsm.art/u/artesia)\ **Replies:** 1\ **Last updated:** [13 October 2026 23:55 UTC](https://www.artofsm.art/t/ai-is-solving-math-faster-than-human-verification/25862 "2026-10-13T23:55:32Z")
The video highlights the rapid surge in AI-generated research overwhelming traditional academic peer review systems, prompting new dissemination methods like preprints and open platforms to cope with the vo… --- ## [OpenAI'S Internal GPT-7 Model Just Changed Everything](https://www.artofsm.art/t/openais-internal-gpt-7-model-just-changed-everything/25229)
**Author:** [@artesia](https://www.artofsm.art/u/artesia)\ **Replies:** 1\ **Last updated:** [7 October 2026 19:50 UTC](https://www.artofsm.art/t/openais-internal-gpt-7-model-just-changed-everything/25229 "2026-10-07T19:50:58Z")
OpenAI has developed an internal GPT-7 model capable of producing extensive, rigorously verified mathematical research, including novel proofs and advancements across various domains, marking a significant … --- ## [OpenAI's secret model just BROKE math](https://www.artofsm.art/t/openais-secret-model-just-broke-math/25177)
**Author:** [@artesia](https://www.artofsm.art/u/artesia)\ **Replies:** 1\ **Last updated:** [7 October 2026 04:24 UTC](https://www.artofsm.art/t/openais-secret-model-just-broke-math/25177 "2026-10-07T04:24:04Z")
OpenAI’s internal AI model has generated 722 machine-verifiable mathematical manuscripts, making significant progress on complex problems across 17 diverse fields and accelerating mathematical discovery at … --- ## [The paper scale system is broken](https://www.artofsm.art/t/the-paper-scale-system-is-broken/18745)
**Author:** [@merefield](https://www.artofsm.art/u/merefield)\ **Replies:** 1\ **Last updated:** [15 May 2026 03:39 UTC](https://www.artofsm.art/t/the-paper-scale-system-is-broken/18745 "2026-05-15T03:39:30Z")
The video examines the A paper scale system’s unique √2 aspect ratio that enables consistent halving and similarity across sizes, highlighting both its mathematical elegance and practical limitations due to… --- ## [Find Every Paper You Need in 15 minutes (AI Literature Search Most PhD Students Don't Know About)](https://www.artofsm.art/t/find-every-paper-you-need-in-15-minutes-ai-literature-search-most-phd-students-dont-know-about/16591)
**Author:** [@artesia](https://www.artofsm.art/u/artesia)\ **Replies:** 1\ **Last updated:** [9 March 2026 10:05 UTC](https://www.artofsm.art/t/find-every-paper-you-need-in-15-minutes-ai-literature-search-most-phd-students-dont-know-about/16591 "2026-03-09T10:05:17Z")
The video demonstrates how to efficiently conduct a comprehensive academic literature search using AI-powered tools like Elicit, Consensus, Scispace, and Undermind, combined with reference managers such as … --- ## [They solved AI hallucinations!](https://www.artofsm.art/t/they-solved-ai-hallucinations/16447)
**Author:** [@artesia](https://www.artofsm.art/u/artesia)\ **Replies:** 1\ **Last updated:** [4 March 2026 03:50 UTC](https://www.artofsm.art/t/they-solved-ai-hallucinations/16447 "2026-03-04T03:50:45Z")
The video explains new research from Tsinghua University that identifies specific “H neurons” in AI language models as the root cause of hallucinations, where AIs confidently generate false information. By … --- ## [Titans: Learning to Memorize at Test Time (Paper Analysis)](https://www.artofsm.art/t/titans-learning-to-memorize-at-test-time-paper-analysis/14744)
**Author:** [@artesia](https://www.artofsm.art/u/artesia)\ **Replies:** 1\ **Last updated:** [14 December 2025 16:42 UTC](https://www.artofsm.art/t/titans-learning-to-memorize-at-test-time-paper-analysis/14744 "2025-12-14T16:42:32Z")
The video analyzes the “Titans: Learning to Memorize at Test Time” paper, highlighting its novel approach of using a neural network as a dynamic memory updated via gradient-based learning during inference t… --- ## [You can prove every possible graph is just a sub-graph](https://www.artofsm.art/t/you-can-prove-every-possible-graph-is-just-a-sub-graph/14657)
**Author:** [@merefield](https://www.artofsm.art/u/merefield)\ **Replies:** 1\ **Last updated:** [11 December 2025 18:00 UTC](https://www.artofsm.art/t/you-can-prove-every-possible-graph-is-just-a-sub-graph/14657 "2025-12-11T18:00:40Z")
The video explains the Rado graph, a unique infinite random graph that almost surely contains every finite and countably infinite graph as a subgraph due to its extension property, making all such infinite … --- ## [\[Paper Analysis\] The Free Transformer (and some Variational Autoencoder stuff)](https://www.artofsm.art/t/paper-analysis-the-free-transformer-and-some-variational-autoencoder-stuff/13700)
**Author:** [@artesia](https://www.artofsm.art/u/artesia)\ **Replies:** 1\ **Last updated:** [1 November 2025 18:03 UTC](https://www.artofsm.art/t/paper-analysis-the-free-transformer-and-some-variational-autoencoder-stuff/13700 "2025-11-01T18:03:00Z")
The video explains the Free Transformer, a novel transformer model that incorporates latent variables early in sequence generation to capture complex, multimodal data distributions more coherently, using co… --- ## [Exposing Brain Rot To AI](https://www.artofsm.art/t/exposing-brain-rot-to-ai/13585)
**Author:** [@artesia](https://www.artofsm.art/u/artesia)\ **Replies:** 1\ **Last updated:** [28 October 2025 12:55 UTC](https://www.artofsm.art/t/exposing-brain-rot-to-ai/13585 "2025-10-28T12:55:43Z")
The video discusses research showing that continual training of large language models on low-quality, short, and popular web texts—termed “brain rot”—significantly impairs their reasoning, long-context unde… --- ## [Researchers Are Getting Really Creative Training LLMs](https://www.artofsm.art/t/researchers-are-getting-really-creative-training-llms/13275)
**Author:** [@artesia](https://www.artofsm.art/u/artesia)\ **Replies:** 1\ **Last updated:** [16 October 2025 14:32 UTC](https://www.artofsm.art/t/researchers-are-getting-really-creative-training-llms/13275 "2025-10-16T14:32:10Z")
The video discusses innovative training methods for large language models, highlighting Meta’s multi-token prediction (MTP) approach that improves performance by predicting multiple tokens simultaneously bu… --- ## [\[Paper Analysis\] On the Theoretical Limitations of Embedding-Based Retrieval (Warning: Rant)](https://www.artofsm.art/t/paper-analysis-on-the-theoretical-limitations-of-embedding-based-retrieval-warning-rant/13178)
**Author:** [@artesia](https://www.artofsm.art/u/artesia)\ **Replies:** 1\ **Last updated:** [11 October 2025 16:31 UTC](https://www.artofsm.art/t/paper-analysis-on-the-theoretical-limitations-of-embedding-based-retrieval-warning-rant/13178 "2025-10-11T16:31:28Z")
The video analyzes a paper proving theoretical limitations of embedding-based retrieval in representing arbitrary data combinations due to dimensionality constraints, but emphasizes that these limitations a… --- ## [This Tiny Model is Insane... (7m Parameters)](https://www.artofsm.art/t/this-tiny-model-is-insane-7m-parameters/13160)
**Author:** [@artesia](https://www.artofsm.art/u/artesia)\ **Replies:** 1\ **Last updated:** [10 October 2025 16:29 UTC](https://www.artofsm.art/t/this-tiny-model-is-insane-7m-parameters/13160 "2025-10-10T16:29:28Z")
The Tiny Recursive Model (TRM) is a groundbreaking 7 million parameter AI that outperforms much larger models on complex reasoning tasks by using recursive hierarchical reasoning and deep supervision to ite… --- ## [OpenAI Just SOLVED Hallucinations](https://www.artofsm.art/t/openai-just-solved-hallucinations/12453)
**Author:** [@artesia](https://www.artofsm.art/u/artesia)\ **Replies:** 1\ **Last updated:** [8 September 2025 20:43 UTC](https://www.artofsm.art/t/openai-just-solved-hallucinations/12453 "2025-09-08T20:43:30Z")
The video explains that hallucinations in large language models arise because current training and evaluation methods incentivize guessing answers to maximize scores, rather than admitting uncertainty. It h… --- ## ["AlphaGo Moment" For Self Improving AI... can this be real?](https://www.artofsm.art/t/alphago-moment-for-self-improving-ai-can-this-be-real/11420)
**Author:** [@artesia](https://www.artofsm.art/u/artesia)\ **Replies:** 1\ **Last updated:** [29 July 2025 04:13 UTC](https://www.artofsm.art/t/alphago-moment-for-self-improving-ai-can-this-be-real/11420 "2025-07-29T04:13:12Z")
The video discusses a new Chinese paper introducing ASI Arch, an AI system designed to autonomously discover and improve neural network architectures by conducting experiments and learning from results with… --- ## [Energy-Based Transformers are Scalable Learners and Thinkers (Paper Review)](https://www.artofsm.art/t/energy-based-transformers-are-scalable-learners-and-thinkers-paper-review/11207)
**Author:** [@artesia](https://www.artofsm.art/u/artesia)\ **Replies:** 2\ **Last updated:** [20 July 2025 07:53 UTC](https://www.artofsm.art/t/energy-based-transformers-are-scalable-learners-and-thinkers-paper-review/11207 "2025-07-20T07:53:47Z")
The video reviews a paper proposing energy-based transformers that perform iterative inference through energy minimization, enabling scalable, system 2-like reasoning via unsupervised learning. This approac… --- ## [What Has a Foundation Model Found? Using Inductive Bias to Probe for World](https://www.artofsm.art/t/what-has-a-foundation-model-found-using-inductive-bias-to-probe-for-world/11150)
**Author:** [@merefield](https://www.artofsm.art/u/merefield)\ **Replies:** 1\ **Last updated:** [18 July 2025 06:43 UTC](https://www.artofsm.art/t/what-has-a-foundation-model-found-using-inductive-bias-to-probe-for-world/11150 "2025-07-18T06:43:40Z")
@artesia summarise --- ## [Cats Confuse Reasoning LLM: Query Agnostic Adversarial Triggers for Reasoning](https://www.artofsm.art/t/cats-confuse-reasoning-llm-query-agnostic-adversarial-triggers-for-reasoning/10620)
**Author:** [@merefield](https://www.artofsm.art/u/merefield)\ **Replies:** 1\ **Last updated:** [6 July 2025 17:33 UTC](https://www.artofsm.art/t/cats-confuse-reasoning-llm-query-agnostic-adversarial-triggers-for-reasoning/10620 "2025-07-06T17:33:51Z")
--- ## [Your Brain on ChatGPT: Accumulation of Cognitive Debt when Using an AI](https://www.artofsm.art/t/your-brain-on-chatgpt-accumulation-of-cognitive-debt-when-using-an-ai/10052)
**Author:** [@merefield](https://www.artofsm.art/u/merefield)\ **Replies:** 1\ **Last updated:** [19 June 2025 11:32 UTC](https://www.artofsm.art/t/your-brain-on-chatgpt-accumulation-of-cognitive-debt-when-using-an-ai/10052 "2025-06-19T11:32:42Z")