AI News Today

arXiv:2505.23866v2 Announce Type: replace Abstract: Deep neural networks have been increasingly used in safety-critical applications such as medical diagnosis and autonomous driving. However, many studies suggest that they are prone to being poorly calibrated and have a propensity for overconfidence, which may have disastrous consequences. In this paper, unlike standard training such as stochastic gradient descent, we show that the recently proposed sharpness-aware minimization (SAM) counteracts

arXiv:2507.20068v3 Announce Type: replace Abstract: Off-policy evaluation (OPE) methods estimate the value of a new reinforcement learning (RL) policy prior to deployment. Recent advances have shown that leveraging auxiliary datasets, such as those synthesized by generative models, can improve the accuracy of OPE methods. Unfortunately, such auxiliary datasets may also be biased, and existing methods for using data augmentation within OPE lack principled uncertainty quantification. In high stake

arXiv:2508.07952v2 Announce Type: replace Abstract: Clustering algorithms often assume all features contribute equally to the data structure, an assumption that usually fails in high-dimensional or noisy settings. Feature weighting methods can address this, but most require additional parameter tuning. We propose SHARK (Shapley Reweighted $k$-means), a feature-weighted clustering algorithm motivated by the use of Shapley values from cooperative game theory to quantify feature relevance, which re

arXiv:2509.14274v3 Announce Type: replace Abstract: Large Language Models (LLMs) have demonstrated significant promise in formal theorem proving. In this study, we investigate the ability of LLMs to discover novel theorems and produce verified proofs. We propose a pipeline called Conjecturing-Proving Loop (CPL), which iteratively generates mathematical conjectures and attempts to prove them in Lean 4. A key feature of CPL is that each iteration conditions the LLM on previously generated theorems

arXiv:2510.01163v2 Announce Type: replace Abstract: The factors driving the performance of in-context learning (ICL) in large language models (LLMs) remain poorly understood despite ICL's surprising effectiveness, enabling models to adapt to new tasks from only a handful of examples. To clarify and improve these capabilities, we characterize how the statistical properties of the pretraining distribution (e.g., tail behavior, coverage) shape ICL. We develop a theoretical framework that encompasse

arXiv:2510.04773v2 Announce Type: replace Abstract: As Large Language Models (LLMs) demonstrate remarkable capabilities learned from vast corpora, concerns regarding data privacy and safety are receiving increasing attention. LLM unlearning, which aims to remove the influence of specific data while preserving overall model utility, is becoming an important research area. One of the mainstream unlearning classes is optimization-based methods, which achieve forgetting directly through fine-tuning,

arXiv:2510.17059v2 Announce Type: replace Abstract: Zero-shot imitation learning requires an agent to reproduce expert behavior from a single demonstration without additional environment interaction or gradient updates at test time. We introduce Contrastive Inverse Reinforcement Learning (CIRL), a self-supervised framework for pre-training zero-shot imitation agents. Our methods rests on a key observation that many useful tasks can be summarized by a single goal state. We can thus convert the mu

arXiv:2512.01759v3 Announce Type: replace Abstract: We investigate the potential of weights to serve as effective representations, focusing on neural fields. Our key insight is that constraining the optimization space through a pre-trained base model and low-rank adaptation (LoRA) can induce structure in weight space. Across reconstruction, generation, and analysis tasks on 2D and 3D data, we find that multiplicative LoRA weights achieve high representation quality while exhibiting distinctivene

arXiv:2512.06244v2 Announce Type: replace Abstract: The exploration-exploitation dilemma in reinforcement learning (RL) is a fundamental challenge to efficient RL algorithms. Existing algorithms for finite state and action discounted RL problems address this by assuming sufficient exploration over both state and action spaces. However, this yields non-implementable algorithms and sub-optimal performance. To resolve these limitations, we introduce a new class of methods with auto-exploration, or

arXiv:2512.10187v3 Announce Type: replace Abstract: LLMs excel at reasoning, but validating their steps remains challenging. Formal verification offers a solution through mechanically checkable proofs. Interactive theorem provers (ITPs) dominate mathematical reasoning but require detailed low-level proof steps, while auto-active verifiers offer automation but focus on software verification. Recent work has begun bridging this divide by evaluating LLMs for software verification in ITPs, but the c

arXiv:2601.02193v2 Announce Type: replace Abstract: We study the extent to which standard machine learning algorithms rely on exchangeability and independence of data by introducing a monotone adversarial corruption model. In this model, an adversary, upon looking at a "clean" i.i.d. dataset, inserts additional "corrupted" points of their choice into the dataset. These added points are constrained to be monotone corruptions, in that they get labeled according to the ground-truth target function.

arXiv:2602.01903v3 Announce Type: replace Abstract: This work studies online episodic tabular Markov decision processes (MDPs) with known transitions and develops best-of-both-worlds algorithms that achieve refined data-dependent regret bounds in the adversarial regime and variance-dependent regret bounds in the stochastic regime. We quantify MDP complexity using a first-order quantity and several new data-dependent measures for the adversarial regime, including a second-order quantity and a pat

arXiv:2602.02877v2 Announce Type: replace Abstract: This paper studies optimization for a family of problems termed $\textbf{compositional entropic risk minimization}$, in which each data's loss is formulated as a Log-Expectation-Exponential (Log-E-Exp) function. The Log-E-Exp formulation serves as an abstraction of the Log-Sum-Exponential (LogSumExp) function when the explicit summation inside the logarithm is taken over a gigantic number of items and is therefore expensive to evaluate. While e

arXiv:2602.05704v2 Announce Type: replace Abstract: Understanding the limitations of gradient methods, and stochastic gradient descent (SGD) in particular, is a central challenge in learning theory. To that end, a commonly used tool is the Statistical Queries (SQ) framework, which studies performance limits of algorithms based on noisy interaction with the data. However, it is known that the formal connection between the SQ framework and SGD is tenuous: Existing results typically rely on adversa

arXiv:2602.06020v3 Announce Type: replace Abstract: How do protein structure prediction models fold proteins? We investigate this question through causal interventions on the folding trunks of ESMFold, OpenFold, and Boltz-1. Across all three models, we find a shared two-stage computational structure. In the first stage, early blocks initialize pairwise biochemical signals: features like charge propagate from sequence into pairwise representations through architecture-specific pathways. In the se

Article URL: https://github.com/surajkumar811/swarm-test Comments URL: https://news.ycombinator.com/item?id=48668678 Points: 1 # Comments: 0

<p> AI memory that connects the dots across every tool </p> <p> <a href="https://www.producthunt.com/products/second-brain-cloudflare?utm_campaign=producthunt-atom-posts-feed&utm_medium=rss-feed&utm_source=producthunt-atom-posts-feed">Discussion</a> | <a href="https://www.producthunt.com/r/p/1180472?app_id=339">Link</a> </p>

Article URL: https://www.forbes.com/sites/jamesbroughel/2026/06/22/bernie-sanders-wants-a-us-sovereign-wealth-fund-for-ai/ Comments URL: https://news.ycombinator.com/item?id=48668581 Points: 4 # Comments: 2

Article URL: https://arxiv.org/abs/2604.07709 Comments URL: https://news.ycombinator.com/item?id=48668552 Points: 2 # Comments: 0

Article URL: https://github.com/Tako-Research/TakoQA Comments URL: https://news.ycombinator.com/item?id=48668549 Points: 1 # Comments: 0

<!-- SC_OFF --><div class="md"><p>Hi, </p> <p>I <strong>really really</strong> need access to Xperience-10M for a deadline which is very soon. </p> <p><a href="https://huggingface.co/datasets/ropedia-ai/xperience-10m">https://huggingface.co/datasets/ropedia-ai/xperience-10m</a></p> <p>Unfortunately, it looks like the owners have stopped approving people to download the dataset. I filled out the form a few weeks ago but have heard nothing back. Several others have also commented on the HF saying

The moment i saw the first llm, i knew the future of tech is keyboardless, actually i was trying to get there before gpt but failed. The fact that its 2026 and most of the industry doesn't see this is baffling to me. When both hardware and software giants are innovating in the same direction (neuralink, apple vision, voice models...etc), You all think using a keyboard and a mouse all day is normal for primates - but it's completely unnatural! it's killing us slowly, both physically & socially. L

Article URL: https://blog.tacoda.dev/this-ones-not-ai-992c95537790 Comments URL: https://news.ycombinator.com/item?id=48668240 Points: 2 # Comments: 2

I had been running a small (3 people) software company for about 4 years. Since closing down, I recently hung out at a friend's company to see what they were working on (15 ppl). To preface: I'm a heavy user of Claude (rarely write code by hand), but what I'm seeing in person has been rather shocking to me, and I wanted to calibrate with others. In particular: - the code is not the source of truth anymore; it's ask claude to write, and ask claude to explain - LoC, abstractions, and all those "so

Article URL: https://github.com/leeguooooo/chrome-use Comments URL: https://news.ycombinator.com/item?id=48668131 Points: 1 # Comments: 0

<pre style='white-space:pre-wrap;width:81ex'>fix(agents): preserve absent embedded session keys</pre>

Article URL: https://tronbrowser.dev/ Comments URL: https://news.ycombinator.com/item?id=48667923 Points: 1 # Comments: 1

A new OpenAI research paper shows how AI agents are transforming work, enabling longer, more complex tasks and expanding productivity across roles.

Article URL: https://atkatana.com/blog/hyper-loops-draft.html Comments URL: https://news.ycombinator.com/item?id=48667742 Points: 1 # Comments: 0

<p> Safe agent spend from the terminal </p> <p> <a href="https://www.producthunt.com/products/paybond?utm_campaign=producthunt-atom-posts-feed&utm_medium=rss-feed&utm_source=producthunt-atom-posts-feed">Discussion</a> | <a href="https://www.producthunt.com/r/p/1180424?app_id=339">Link</a> </p>

A major challenge to what we thought. ScienceAlert stories are written, fact-checked, and edited by humans, never generated by AI. Don't miss a story, subscribe here.

<pre style='white-space:pre-wrap;width:81ex'>fix(cron): preserve enabled-with-defaults failure alert through store roundtrip (fixes #96589) (AI-assisted) (#96615) Summary: - The PR preserves `failure_alert_disabled === 0` as the enabled-with-defaults failure-alert state and adds focused codec roundtrip tests. - PR surface: Source +2, Tests +54. Total +56 across 2 files. - Reproducibility: yes. At source level, current main encodes `failureAlert: {}` with `failure_alert_disabled = 0`, then decode

Article URL: https://github.com/naya-ai/promptctl Comments URL: https://news.ycombinator.com/item?id=48667544 Points: 2 # Comments: 0

Here's how the new Google Home Speaker has fared against the likes of my Apple HomePod Mini and Echo Dot Max so far.

Article URL: https://www.jimmont.com/llm-style-token-costs Comments URL: https://news.ycombinator.com/item?id=48667409 Points: 3 # Comments: 2

Article URL: https://github.com/Kormiloio/Halyard Comments URL: https://news.ycombinator.com/item?id=48667366 Points: 1 # Comments: 0

Save $200 on MSI's Katana 15 HX during Prime Day. It houses an Intel Core i7 CPU, RTX 5060 GPU, and a 165Hz QHD display.

<pre style='white-space:pre-wrap;width:81ex'>fix(agent): emit model.usage diagnostic for HTTP ingress traffic (#96152) * fix(agents): emit model.usage diagnostic for HTTP ingress traffic * fix(agents): emit model.usage diagnostic for HTTP ingress traffic * fix(agent): add regression tests and refactor ingress model.usage diagnostic emission * fix(agent): resolve oxlint curly and no-useless-fallback-in-spread violations</pre>

As ASML CEO Christophe Fouquet told TechCrunch in May, what China can currently buy are older-generation deep ultraviolet tools — gear first shipped about a decade ago — the same machines the MATCH Act would now put off limits.

Article URL: https://www.anildash.com/2026/06/23/fight-ai-platform-war/ Comments URL: https://news.ycombinator.com/item?id=48667112 Points: 3 # Comments: 0