https://icymilaw.org/#aipapers Popular AI paper-like links shared recently on Bluesky. en-us Wed, 12 Aug 2026 06:10:20 -0400 https://arxiv.org/abs/2608.07009 https://arxiv.org/abs/2608.07009 <p><img src="https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:nrfqk6t446qme4kb7aiemqig/bafkreideav22at7bx4sm5tyeqbxcidb4fmal7ir2pfpzlxzpgnmmlzufbm" alt="" style="max-width:100%;height:auto;" /></p>ArXiv link for HiSparse: Scaling Sparse-Attention Decoding with Hierarchical KV Cache Management<br><br><small> 🔥 · reposts: 1 · likes: 13</small> Wed, 12 Aug 2026 06:10:20 -0400 arxiv.org https://arxiv.org/abs/2608.06931 https://arxiv.org/abs/2608.06931 <p><img src="https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:nrfqk6t446qme4kb7aiemqig/bafkreigwru6vhlzzw3pfun2tn5k3f2t6w6vtcrednwom2tqpxepq3ktbxa" alt="" style="max-width:100%;height:auto;" /></p>ArXiv link for Science Edge Evaluation: SEE the Missing Step Toward Real Scientific Discovery<br><br><small> 🔥 · reposts: 1 · likes: 7</small> Wed, 12 Aug 2026 06:10:20 -0400 arxiv.org https://arxiv.org/abs/2604.21827 https://arxiv.org/abs/2604.21827 <p><img src="https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:nrfqk6t446qme4kb7aiemqig/bafkreifbdvy3zuikvktgozavmabzdj4gwbeayqxprcct3cedzfw52mhrx4" alt="" style="max-width:100%;height:auto;" /></p>ArXiv link for Alignment has a Fantasia Problem<br><br><small>reposts: 1 · likes: 4</small> Wed, 12 Aug 2026 06:10:20 -0400 arxiv.org https://arxiv.org/abs/2505.16170 https://arxiv.org/abs/2505.16170 <p><img src="https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:nrfqk6t446qme4kb7aiemqig/bafkreiajort52gifn3unzdvk6fgjtcwnyaryvr4ysm5uok5yevzmrpqihy" alt="" style="max-width:100%;height:auto;" /></p>ArXiv link for When Do LLMs Admit Their Mistakes? Understanding The Role Of Model Belief In Retraction<br><br><small>reposts: 1 · likes: 4</small> Wed, 12 Aug 2026 06:10:20 -0400 arxiv.org https://arxiv.org/abs/2105.06948 https://arxiv.org/abs/2105.06948 <p><img src="https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:5j3itjwkgio4elgmgtxqgtjb/bafkreigc4ofbn4qhcdzeaiy7q2h4qsynlyzvuzqcxetqfipq4bamvd2n2i" alt="" style="max-width:100%;height:auto;" /></p>One of the most striking features of human cognition is the capacity to plan. Two aspects of human planning stand out: its efficiency and flexibility. Efficiency is especially impressive because plans...<br><br><small> 🔥🔥 · reposts: 3 · likes: 20</small> Mon, 10 Aug 2026 06:08:11 -0400 arxiv.org https://arxiv.org/abs/2510.09022 https://arxiv.org/abs/2510.09022 <p><img src="https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:zfo3ww2suaikfjuho3lvs7zs/bafkreid5ef6vsscqwei75f6d6ixwxru7unutp2iqw6zabq2c7k3hykpboq" alt="" style="max-width:100%;height:auto;" /></p>Recent Machine Learning (ML) approaches have shown increased performance on benchmarks but at the cost of escalating computational demands. Hardware, algorithmic and carbon optimizations have been pro...<br><br><small> 🔥🔥🔥 · reposts: 4 · likes: 19</small> Sun, 09 Aug 2026 06:09:09 -0400 arxiv.org https://arxiv.org/abs/2608.02859 https://arxiv.org/abs/2608.02859 <p><img src="https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:7bcjnl2b5ko25d5zgdpiqgop/bafkreifa2wylk6yt7yvshmvvwedptaubw7glvquzoen5u5xnqmu7isayhm" alt="" style="max-width:100%;height:auto;" /></p>In this essay, I present the case for total opposition to the use of artificial intelligence in mathematics. I offer proposals for how individuals, departments, journals, and institutions can act in c...<br><br><small> 🔥 · reposts: 2 · likes: 9</small> Sun, 09 Aug 2026 06:09:09 -0400 arxiv.org https://arxiv.org/abs/2608.05245 https://arxiv.org/abs/2608.05245 <p><img src="https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:nrfqk6t446qme4kb7aiemqig/bafkreidl7iykvibbmybf56255o6ozyu3qbvqaa3vlc63m4i54plbuyj5lu" alt="" style="max-width:100%;height:auto;" /></p>ArXiv link for Search2Skill: Skill Distillation Beyond Knowledge Boundaries Via Rubric-Based Reinforcement Learning<br><br><small> 🔥 · reposts: 3 · likes: 5</small> Sun, 09 Aug 2026 06:09:09 -0400 arxiv.org https://arxiv.org/abs/2607.18056 https://arxiv.org/abs/2607.18056 <p><img src="https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:3wdc7dmw4l7uxgwfwixm7lc5/bafkreigc4ofbn4qhcdzeaiy7q2h4qsynlyzvuzqcxetqfipq4bamvd2n2i" alt="" style="max-width:100%;height:auto;" /></p>Frontier large language models (LLMs) are increasingly integrated into scientific workflows, yet their growing biological capabilities may outpace current safeguards. To assess the biological risks of...<br><br><small> 🔥🔥 · reposts: 4 · likes: 8</small> Sat, 08 Aug 2026 06:10:08 -0400 arxiv.org https://arxiv.org/abs/2603.12471 https://arxiv.org/abs/2603.12471 <p><img src="https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:nmoaombfcftmgt6emgcapyh5/bafkreibmwume2q2jvfyf5jvz62lbuyuyxmwh2ewnlggfcjyro263dps63m" alt="" style="max-width:100%;height:auto;" /></p>Effective personalized feedback is critical to students&#x27; literacy development. Though LLM-powered tools now promise to automate such feedback at scale, LLMs are not language-neutral: they privilege st...<br><br><small> 🔥 · reposts: 3 · likes: 4</small> Sat, 08 Aug 2026 06:10:08 -0400 arxiv.org https://arxiv.org/abs/2608.02642 https://arxiv.org/abs/2608.02642 <p><img src="https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:lykmonumufb32zwuv25rfnku/bafkreifa2wylk6yt7yvshmvvwedptaubw7glvquzoen5u5xnqmu7isayhm" alt="" style="max-width:100%;height:auto;" /></p>Accelerating scientific discovery is among the most consequential applications of AI, and computational biomolecular simulation stands out as a particularly promising target within this broader effort...<br><br><small> 🔥 · reposts: 5 · likes: 13</small> Thu, 06 Aug 2026 06:09:55 -0400 arxiv.org https://papers.ssrn.com/sol3/papers.cfm?abstract_id=7234698 https://papers.ssrn.com/sol3/papers.cfm?abstract_id=7234698 Amici curiae are dozens of professors of law, economics, and business who have researched and published extensively on antitrust issue<br><br><small> 🔥 · reposts: 6 · likes: 9</small> Thu, 06 Aug 2026 06:09:55 -0400 papers.ssrn.com https://arxiv.org/abs/2607.13560 https://arxiv.org/abs/2607.13560 <p><img src="https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:llojyojekfb3vzxl44txm65v/bafkreidqmqwy3yg4tl3phag6b33z4zqxmqu4b2m6eya4gutzr2weqdsdui" alt="" style="max-width:100%;height:auto;" /></p>Recent advances in generative and embodied AI have been driven by large-scale predictive learning over multimodal data. However, the resulting systems remain largely based on passive training regimes ...<br><br><small> 🔥 · reposts: 3 · likes: 14</small> Thu, 06 Aug 2026 06:09:55 -0400 arxiv.org https://arxiv.org/abs/2401.11817 https://arxiv.org/abs/2401.11817 <p><img src="https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:zfo3ww2suaikfjuho3lvs7zs/bafkreid5ef6vsscqwei75f6d6ixwxru7unutp2iqw6zabq2c7k3hykpboq" alt="" style="max-width:100%;height:auto;" /></p>Hallucination has been widely recognized to be a significant drawback for large language models (LLMs). There have been many works that attempt to reduce the extent of hallucination. These efforts hav...<br><br><small> 🔥🔥🔥 · reposts: 19 · likes: 47</small> Wed, 05 Aug 2026 06:11:54 -0400 arxiv.org https://arxiv.org/abs/2607.25166 https://arxiv.org/abs/2607.25166 <p><img src="https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:eheolvrmn7vchwe3n5bclsej/bafkreid2igg2nszsqfcgmgodcyiwambgidqap7cyjdwjqqwhk5dr2i4hsa" alt="" style="max-width:100%;height:auto;" /></p>AI chatbots can be &quot;sycophantic,&quot; or overly agreeable and flattering toward users. Sycophantic AI has been shown to entrench attitudes, yet users frequently fail to recognize it (a phenomenon we call ...<br><br><small> 🔥 · reposts: 12 · likes: 17</small> Wed, 05 Aug 2026 06:11:54 -0400 arxiv.org https://arxiv.org/abs/2601.05047 https://arxiv.org/abs/2601.05047 <p><img src="/static/browse/0.3.4/images/arxiv-logo-fb.png" alt="" style="max-width:100%;height:auto;" /></p>Large Language Model (LLM) inference is hard. The autoregressive Decode phase of the underlying Transformer model makes LLM inference fundamentally different from training. Exacerbated by recent AI tr...<br><br><small> 🔥 · reposts: 6 · likes: 24</small> Wed, 05 Aug 2026 06:11:54 -0400 arxiv.org https://arxiv.org/abs/2312.10794 https://arxiv.org/abs/2312.10794 <p><img src="https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:tnidwtbadagrgztdyrwz7rtu/bafkreigc4ofbn4qhcdzeaiy7q2h4qsynlyzvuzqcxetqfipq4bamvd2n2i" alt="" style="max-width:100%;height:auto;" /></p>Transformers play a central role in the inner workings of large language models. We develop a mathematical framework for analyzing Transformers based on their interpretation as interacting particle sy...<br><br><small>reposts: 0 · likes: 12</small> Wed, 05 Aug 2026 06:11:54 -0400 arxiv.org https://arxiv.org/abs/2607.27191 https://arxiv.org/abs/2607.27191 <p><img src="https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:3igkkqe5nkm66olc4y2syzpy/bafkreigc4ofbn4qhcdzeaiy7q2h4qsynlyzvuzqcxetqfipq4bamvd2n2i" alt="" style="max-width:100%;height:auto;" /></p>Forecasts of explosive AI progress hinge on AI agents automating AI research. But evidence on whether agents can carry out open-ended AI research is thin. Current evaluations either test agents on nar...<br><br><small> 🔥🔥🔥 · reposts: 23 · likes: 95</small> Tue, 04 Aug 2026 06:09:31 -0400 arxiv.org https://arxiv.org/abs/2507.01231 https://arxiv.org/abs/2507.01231 <p><img src="https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:zfo3ww2suaikfjuho3lvs7zs/bafkreid5ef6vsscqwei75f6d6ixwxru7unutp2iqw6zabq2c7k3hykpboq" alt="" style="max-width:100%;height:auto;" /></p>Earlier this year, Apple ignited controversy by publishing &quot;The Illusion of Thinking,&quot; prompting heated debate within the AI community. Critics seized upon the findings as conclusive evidence that Lar...<br><br><small>reposts: 4 · likes: 13</small> Mon, 03 Aug 2026 06:08:29 -0400 arxiv.org https://arxiv.org/abs/2603.18161 https://arxiv.org/abs/2603.18161 <p><img src="https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:5wm25vgenhgut3iqfjf4ozj5/bafkreigc4ofbn4qhcdzeaiy7q2h4qsynlyzvuzqcxetqfipq4bamvd2n2i" alt="" style="max-width:100%;height:auto;" /></p>Large language models (LLMs) are used by over a billion people globally, most often to assist with writing. In this work, we demonstrate that LLMs not only alter the voice and tone of human writing, b...<br><br><small> 🔥🔥🔥🔥🔥🔥 · reposts: 44 · likes: 146</small> Sat, 01 Aug 2026 06:08:39 -0400 arxiv.org https://arxiv.org/abs/2603.12277 https://arxiv.org/abs/2603.12277 <p><img src="https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:p72jkscm4rt4xuep3lzv6doh/bafkreigc4ofbn4qhcdzeaiy7q2h4qsynlyzvuzqcxetqfipq4bamvd2n2i" alt="" style="max-width:100%;height:auto;" /></p>LLMs see the world as a single stream of text, partitioned into roles like or . We trace prompt injection to role confusion: models perceive the source of text from how it sounds, not its...<br><br><small> 🔥🔥 · reposts: 16 · likes: 60</small> Sat, 01 Aug 2026 06:08:39 -0400 arxiv.org https://arxiv.org/abs/2607.27978 https://arxiv.org/abs/2607.27978 <p><img src="https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:nqqmph3m6umv5p5jlmudro3k/bafkreifa2wylk6yt7yvshmvvwedptaubw7glvquzoen5u5xnqmu7isayhm" alt="" style="max-width:100%;height:auto;" /></p>In spring 2026, an economics professor at Brown University gave a take-home midterm and, after unusually high scores, made the final exam proctored. Among the 59 students who completed the course, ave...<br><br><small> 🔥 · reposts: 14 · likes: 37</small> Sat, 01 Aug 2026 06:08:39 -0400 arxiv.org https://arxiv.org/abs/2607.28227 https://arxiv.org/abs/2607.28227 <p><img src="https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:27ivzcszryxp6mehutodmcxo/bafkreicl7bxvv7xxihrk4m2vj6tfjlnrpqlw4yr6asfl4z3olalfmtlhe4" alt="" style="max-width:100%;height:auto;" /></p>GUI agents have the potential to become a general purpose executor over existing digital devices. To advance them toward real-world use, we envision agents that operate reliably on real devices, execu...<br><br><small> 🔥 · reposts: 7 · likes: 30</small> Sat, 01 Aug 2026 06:08:39 -0400 arxiv.org https://arxiv.org/abs/2607.17397 https://arxiv.org/abs/2607.17397 <p><img src="https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:sgzwcbgta3lha4gmnwvkgvbb/bafkreibmwume2q2jvfyf5jvz62lbuyuyxmwh2ewnlggfcjyro263dps63m" alt="" style="max-width:100%;height:auto;" /></p>As a labor-augmenting technology, large language models (LLMs) have the potential to accelerate scientific activity across the research pipeline. But even if LLMs perform on par with human experts at ...<br><br><small> 🔥 · reposts: 4 · likes: 27</small> Sat, 01 Aug 2026 06:08:39 -0400 arxiv.org https://arxiv.org/abs/2607.27197 https://arxiv.org/abs/2607.27197 <p><img src="https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:yzywgiiou7cx63uddiru6m2o/bafkreidqmqwy3yg4tl3phag6b33z4zqxmqu4b2m6eya4gutzr2weqdsdui" alt="" style="max-width:100%;height:auto;" /></p>We exhibit a configuration of five point charges in Euclidean space whose electrostatic potential admits at least 24 critical points all of which are non-degenerate. Maxwell&#x27;s conjecture that the fiel...<br><br><small> 🔥🔥🔥🔥 · reposts: 2 · likes: 24</small> Fri, 31 Jul 2026 06:09:27 -0400 arxiv.org https://papers.ssrn.com/sol3/papers.cfm?abstract_id=7192378 https://papers.ssrn.com/sol3/papers.cfm?abstract_id=7192378 Amici curiae are scholars of intellectual property. Amici have no personal interest in the outcome of this litigation but share a professional interest in seein<br><br><small> 🔥 · reposts: 3 · likes: 3</small> Fri, 31 Jul 2026 06:09:27 -0400 papers.ssrn.com https://arxiv.org/abs/2607.25845 https://arxiv.org/abs/2607.25845 <p><img src="https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:nkcgn7kpbsidnhq55gdfnm3a/bafkreigc4ofbn4qhcdzeaiy7q2h4qsynlyzvuzqcxetqfipq4bamvd2n2i" alt="" style="max-width:100%;height:auto;" /></p>Static retrieval frameworks are leading tools for interpreting exoplanet observations, yet time-independent modelling leaves them prone to degeneracy and unable to resolve exoplanets&#x27; histories. The c...<br><br><small> 🔥🔥🔥 · reposts: 8 · likes: 19</small> Thu, 30 Jul 2026 06:10:25 -0400 arxiv.org https://arxiv.org/abs/2607.25398 https://arxiv.org/abs/2607.25398 <p><img src="/static/browse/0.3.4/images/arxiv-logo-fb.png" alt="" style="max-width:100%;height:auto;" /></p>Language-model agents are increasingly deployed under standing instructions: a system prompt, a policy file, or a skills document is placed in context, and the agent is trusted to let it govern every action that follows. Existing benchmarks rarely test this deployment pattern directly; they measure whether an agent can complete a task, not whether a long, binding policy document actually constrains its behavior over an extended tool-use horizon. We present HANDBOOK.md, a benchmark of 65 agentic tasks modeled on how enterprise employees follow company handbooks. Each task places an agent in a self-contained company environment, a file workspace together with mock email, chat, calendar, issue-tracking, and commerce services exposed over the Model Context Protocol, and instructs it to carry out routine professional work governed by an expert-written standard operating procedure of 20 to 124 pages. Tasks span five domains (finance, medical billing, insurance, logistics, and HR) and ten fictional companies. To resist memorization, every task modifies one of ten base handbooks, altering the specific rules and thresholds on which grading turns, so no two tasks share a policy. Grading is fully deterministic: each task carries a rubric of programmatic criteria (824 in total) that check both that required actions occurred and that prohibited actions did not. Under strict grading, where a trial passes only if every criterion is satisfied, the best of thirty evaluated model configurations passes 36.2% of trials, and most frontier configurations remain below 25%. Failures follow consistent patterns: agents let a plausible in-environment request override the standing policy, perform a required check and then act against its result, lose rule details over long horizons, and report compliance they did not achieve. We release all tasks, environments, and the evaluation harness.<br><br><small> 🔥 · reposts: 2 · likes: 7</small> Thu, 30 Jul 2026 06:10:25 -0400 arxiv.org https://arxiv.org/abs/2606.14580 https://arxiv.org/abs/2606.14580 <p><img src="https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:6xpwzearwjjlfyb7l2oon2gp/bafkreidqmqwy3yg4tl3phag6b33z4zqxmqu4b2m6eya4gutzr2weqdsdui" alt="" style="max-width:100%;height:auto;" /></p>Identifying persuasive rhetorical cues is critical across domains, from detecting information manipulation and improving AI safety to advancing public health communication. We propose Persuasion Index...<br><br><small> 🔥 · reposts: 1 · likes: 8</small> Wed, 29 Jul 2026 06:10:22 -0400 arxiv.org https://arxiv.org/abs/2607.21616 https://arxiv.org/abs/2607.21616 <p><img src="https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:nrfqk6t446qme4kb7aiemqig/bafkreiacpe7g7hmthno3ll5wksotepeawjcvmpymovbmlrwckzkd53t34y" alt="" style="max-width:100%;height:auto;" /></p>ArXiv link for Lost in Context: Addressing Context Anxiety in Large Language Models<br><br><small> 🔥 · reposts: 2 · likes: 5</small> Wed, 29 Jul 2026 06:10:22 -0400 arxiv.org https://arxiv.org/abs/2607.22681 https://arxiv.org/abs/2607.22681 <p><img src="https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:mnly77ciucowvudu2dw6w56z/bafkreid5ef6vsscqwei75f6d6ixwxru7unutp2iqw6zabq2c7k3hykpboq" alt="" style="max-width:100%;height:auto;" /></p>This report studies three nuances in peer review data: paper version, score version, and input format. We characterize how the variants differ, and measure their impact on downstream tasks. Based on o...<br><br><small>reposts: 1 · likes: 4</small> Wed, 29 Jul 2026 06:10:22 -0400 arxiv.org https://papers.ssrn.com/sol3/papers.cfm?abstract_id=7191580 https://papers.ssrn.com/sol3/papers.cfm?abstract_id=7191580 Large language models (LLMs) are increasingly used to generate data for research. Typical use cases are classifications, annotations, information extraction, an<br><br><small> 🔥🔥 · reposts: 4 · likes: 25</small> Tue, 28 Jul 2026 06:08:35 -0400 papers.ssrn.com https://arxiv.org/abs/2605.26492 https://arxiv.org/abs/2605.26492 <p><img src="https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:nv5oeyb6wfhvqdjniea2mvsu/bafkreidqmqwy3yg4tl3phag6b33z4zqxmqu4b2m6eya4gutzr2weqdsdui" alt="" style="max-width:100%;height:auto;" /></p>LLM-generated stories are a popular use case, but they show very low variability. We sample 20,000 total stories from four current models using five prompts. We find that 11 words occur in 88.3% of ge...<br><br><small> 🔥 · reposts: 2 · likes: 13</small> Tue, 28 Jul 2026 06:08:35 -0400 arxiv.org https://arxiv.org/abs/2607.20695 https://arxiv.org/abs/2607.20695 <p><img src="https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:sgzwcbgta3lha4gmnwvkgvbb/bafkreibmwume2q2jvfyf5jvz62lbuyuyxmwh2ewnlggfcjyro263dps63m" alt="" style="max-width:100%;height:auto;" /></p>Human judgment is fundamentally prone to error. A promise of AI is that it will rid decisions of bias and ensure a fairer and safer world for all. Yet research unequivocally demonstrates that LLMs exh...<br><br><small> 🔥🔥🔥 · reposts: 17 · likes: 32</small> Mon, 27 Jul 2026 06:07:53 -0400 arxiv.org https://arxiv.org/abs/2607.20149 https://arxiv.org/abs/2607.20149 <p><img src="https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:sgzwcbgta3lha4gmnwvkgvbb/bafkreibmwume2q2jvfyf5jvz62lbuyuyxmwh2ewnlggfcjyro263dps63m" alt="" style="max-width:100%;height:auto;" /></p>Machine learning courses often use pre-labeled datasets, hiding the subjectivity of human annotation. This creates students with an overly trusting view of AI data and models, undervaluing interpretiv...<br><br><small> 🔥🔥 · reposts: 9 · likes: 27</small> Mon, 27 Jul 2026 06:07:53 -0400 arxiv.org https://arxiv.org/abs/2607.21534 https://arxiv.org/abs/2607.21534 <p><img src="https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:sgzwcbgta3lha4gmnwvkgvbb/bafkreibmwume2q2jvfyf5jvz62lbuyuyxmwh2ewnlggfcjyro263dps63m" alt="" style="max-width:100%;height:auto;" /></p>The spread of generative AI (GenAI) in higher education has raised concerns that students offload cognitive effort to AI, earning high grades without learning. If this &quot;GenAI substitution hypothesis&quot; ...<br><br><small> 🔥 · reposts: 5 · likes: 13</small> Mon, 27 Jul 2026 06:07:53 -0400 arxiv.org https://arxiv.org/abs/2607.20349 https://arxiv.org/abs/2607.20349 <p><img src="https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:sgzwcbgta3lha4gmnwvkgvbb/bafkreibmwume2q2jvfyf5jvz62lbuyuyxmwh2ewnlggfcjyro263dps63m" alt="" style="max-width:100%;height:auto;" /></p>Generative AI can produce book-length works of fiction at near-zero cost. These books are often dismissed as low-quality ``slop&#x27;&#x27; that buyers will ignore, and are assumed to carry little commercial we...<br><br><small>reposts: 3 · likes: 4</small> Sun, 26 Jul 2026 06:10:03 -0400 arxiv.org https://arxiv.org/abs/2607.20982 https://arxiv.org/abs/2607.20982 <p><img src="https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:nrfqk6t446qme4kb7aiemqig/bafkreibd2vybt5niis7zyp2lmv3fqg6bwk66sxug3kgn2zcb7mhtwkqwyq" alt="" style="max-width:100%;height:auto;" /></p>ArXiv link for GuardianAgentBench: Where Agents Fail and How to Guard Them<br><br><small>reposts: 1 · likes: 4</small> Sun, 26 Jul 2026 06:10:03 -0400 arxiv.org https://arxiv.org/abs/2606.16475 https://arxiv.org/abs/2606.16475 <p><img src="https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:5wm25vgenhgut3iqfjf4ozj5/bafkreifa2wylk6yt7yvshmvvwedptaubw7glvquzoen5u5xnqmu7isayhm" alt="" style="max-width:100%;height:auto;" /></p>Many societal decisions are settled by contests of persuasion. Conversational AI is a powerful new entrant in these contests, but whether it can out-persuade skilled and highly incentivized humans has...<br><br><small> 🔥🔥🔥🔥🔥🔥 · reposts: 31 · likes: 128</small> Sat, 25 Jul 2026 09:09:10 -0400 arxiv.org https://arxiv.org/abs/2607.21595 https://arxiv.org/abs/2607.21595 <p><img src="https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:nrfqk6t446qme4kb7aiemqig/bafkreif56ubxo5euijkzr4jszouftfceppol3ia2tekrqrz2twjhxcriba" alt="" style="max-width:100%;height:auto;" /></p>ArXiv link for 3D-Aware VLMs with Implicit and Explicit Geometries<br><br><small>reposts: 1 · likes: 5</small> Sat, 25 Jul 2026 09:09:10 -0400 arxiv.org https://arxiv.org/abs/2510.26692?utm_campaign=The%20Batch&utm_medium=email&_hsenc=p2ANqtz-8KAuEYp4RL56msN-qThdZgIy3lXiIqGLL6swFd0mZ9ti_Y_h6ryr0UQiNhgWFwcbBe9uDOtYL2ypZq5jWi8QHgnUAdeenf81xGN56C8to9entIIb4&_hsmi=429858740&utm_content=429859766&utm_source=hs_email https://arxiv.org/abs/2510.26692?utm_campaign=The%20Batch&utm_medium=email&_hsenc=p2ANqtz-8KAuEYp4RL56msN-qThdZgIy3lXiIqGLL6swFd0mZ9ti_Y_h6ryr0UQiNhgWFwcbBe9uDOtYL2ypZq5jWi8QHgnUAdeenf81xGN56C8to9entIIb4&_hsmi=429858740&utm_content=429859766&utm_source=hs_email <p><img src="https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:qninmqhyuqoi3qiaipmtqcoy/bafkreifa2wylk6yt7yvshmvvwedptaubw7glvquzoen5u5xnqmu7isayhm" alt="" style="max-width:100%;height:auto;" /></p>We introduce Kimi Linear, a hybrid linear attention architecture that, for the first time, outperforms full attention under fair comparisons across various scenarios -- including short-context, long-c...<br><br><small>reposts: 1 · likes: 2</small> Sat, 25 Jul 2026 09:09:10 -0400 arxiv.org https://arxiv.org/abs/2607.20255 https://arxiv.org/abs/2607.20255 <p><img src="https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:sgzwcbgta3lha4gmnwvkgvbb/bafkreibmwume2q2jvfyf5jvz62lbuyuyxmwh2ewnlggfcjyro263dps63m" alt="" style="max-width:100%;height:auto;" /></p>LLM-driven autonomous agents are reshaping offensive security. Unlike traditional penetration-testing tooling -- deterministic, narrowly scoped, and operated by trained practitioners -- agentic securi...<br><br><small> 🔥 · reposts: 1 · likes: 5</small> Fri, 24 Jul 2026 06:10:38 -0400 arxiv.org http://papers.ssrn.com/sol3/papers.cfm?abstract_id=7151718 http://papers.ssrn.com/sol3/papers.cfm?abstract_id=7151718 AI artifacts like models, datasets, and web platforms increasingly help build other AI artifacts. Developers rely on existing models to generat<br><br><small>reposts: 1 · likes: 3</small> Fri, 24 Jul 2026 06:10:38 -0400 papers.ssrn.com https://arxiv.org/abs/2607.17979 https://arxiv.org/abs/2607.17979 <p><img src="https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:nrfqk6t446qme4kb7aiemqig/bafkreibfbwc3adchyjd7asemi4aoz6qqfyjoaxkdrvwz47i6ctrefmkxwa" alt="" style="max-width:100%;height:auto;" /></p>ArXiv link for Harness Engineering for LLM-Driven GPU Kernel Generation<br><br><small>reposts: 1 · likes: 2</small> Fri, 24 Jul 2026 06:10:38 -0400 arxiv.org https://arxiv.org/abs/2304.15004?hl=de-DE https://arxiv.org/abs/2304.15004?hl=de-DE <p><img src="https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:zfo3ww2suaikfjuho3lvs7zs/bafkreid5ef6vsscqwei75f6d6ixwxru7unutp2iqw6zabq2c7k3hykpboq" alt="" style="max-width:100%;height:auto;" /></p>Recent work claims that large language models display emergent abilities, abilities not present in smaller-scale models that are present in larger-scale models. What makes emergent abilities intriguin...<br><br><small> 🔥🔥🔥🔥🔥 · reposts: 4 · likes: 18</small> Thu, 23 Jul 2026 06:12:08 -0400 arxiv.org https://arxiv.org/abs/2510.20075 https://arxiv.org/abs/2510.20075 <p><img src="https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:6ond5sxlegjxpe3ismrczk3r/bafkreigc4ofbn4qhcdzeaiy7q2h4qsynlyzvuzqcxetqfipq4bamvd2n2i" alt="" style="max-width:100%;height:auto;" /></p>A meaningful text can be hidden inside another, completely different yet still coherent and plausible, text of the same length. For example, a tweet containing a harsh political critique could be embe...<br><br><small> 🔥🔥🔥 · reposts: 1 · likes: 15</small> Thu, 23 Jul 2026 06:12:08 -0400 arxiv.org https://arxiv.org/abs/2404.01549 https://arxiv.org/abs/2404.01549 <p><img src="https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:nrfqk6t446qme4kb7aiemqig/bafkreiaqlol2ufh7wfsvdrv2q45ejuqaxj5mczj5vvjk3qnrrmijg5b6ye" alt="" style="max-width:100%;height:auto;" /></p>ArXiv link for Octopus: On-device language model for function calling of software APIs<br><br><small>reposts: 0 · likes: 4</small> Thu, 23 Jul 2026 06:12:08 -0400 arxiv.org https://arxiv.org/abs/2404.11459 https://arxiv.org/abs/2404.11459 <p><img src="https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:nrfqk6t446qme4kb7aiemqig/bafkreih5b4j2pgvvipl3kfbe4xr3ptp2idpevka5zve4tjxhgmq3hq6cu4" alt="" style="max-width:100%;height:auto;" /></p>ArXiv link for Octopus v3: Technical Report for On-device Sub-billion Multimodal AI Agent<br><br><small> 🔥 · reposts: 1 · likes: 5</small> Wed, 22 Jul 2026 06:11:48 -0400 arxiv.org https://papers.ssrn.com/sol3/papers.cfm?abstract_id=7146879 https://papers.ssrn.com/sol3/papers.cfm?abstract_id=7146879 Legal and policy discussions on the environmental impacts of AI have drawn attention to the industry&#x27;s climate impacts, pollution and resource consumption. In c<br><br><small> 🔥🔥🔥🔥 · reposts: 4 · likes: 12</small> Tue, 21 Jul 2026 06:10:23 -0400 papers.ssrn.com https://arxiv.org/abs/2607.15769 https://arxiv.org/abs/2607.15769 <p><img src="/static/browse/0.3.4/images/arxiv-logo-fb.png" alt="" style="max-width:100%;height:auto;" /></p>Generative AI and coding agents are intensifying a central governance tension in open-source software (OSS): they scale contribution generation faster than maintainers can assess risk, evidence, and accountability. Existing responses improve agent-readability and traceability, but project rules mus…<br><br><small> 🔥 · reposts: 2 · likes: 3</small> Tue, 21 Jul 2026 06:10:23 -0400 arxiv.org