Checking the Reasoning of AI’s Black Boxes
A new framework offers a practical way to verify the behavior of large language models without needing access to their internal workings.
Science
Geometric Codes: New Links Between Curves and Error Correction
Researchers have discovered a novel connection between geometric structures and error-correcting codes, expanding our understanding of both fields.
Building Reliable AI Software with Event History
A new architecture leverages the power of event sourcing to bring greater predictability and auditability to software developed with large language models.
The Text Decoder: Pinpointing How Vision-Language Models ‘Read’ Images
New research reveals specific layers within these models that act as crucial bottlenecks for optical character recognition, offering a pathway to understand and control their visual processing abilities.
Universal Rules for Charm Decay
![The comparison of next-to-leading order (NLO) sum rules-detailed in [latex]Eqs. (65, 71, 84)[/latex]-to their leading order (LO) counterparts-defined by [latex]Eqs. (64, 69, 70, 77, 78, 79, 80)[/latex]-reveals the subtle shifts in understanding as calculations refine, all while remaining bounded by the fundamental constraints illustrated by the UU-spin limit.](https://arxiv.org/html/2602.22320v1/2602.22320v1/x1.png)
New research reveals a surprisingly simple and robust relationship governing the decay of charm particles.
Layered Perovskites: A Delicate Balance of Spin and Orbit
![The ferromagnetic phase exhibits a dependence between the pseudoorbital-space polar angle θ and the variables [latex]J-\lambda[/latex], particularly when [latex]t\ll\Delta_{\rm CF}\ll U[/latex] with [latex]U=20t[/latex], demonstrating a constrained relationship within specific energy scales.](https://arxiv.org/html/2602.23011v1/2602.23011v1/theta2.jpg)
New research explores the complex interplay of interactions governing magnetic behavior in layered perovskite materials.
The Art of the Prompt: Bypassing AI Safety with Ancient Language
Researchers have discovered a surprising method for circumventing AI safety protocols by crafting adversarial prompts rooted in the nuances of Classical Chinese.
Unlocking the Rho Meson: A New Path to Precision Calculations
Researchers are combining effective field theory and lattice QCD to develop a rigorous framework for calculating the electromagnetic properties of the rho meson.
Building Artificial Minds: A New Loop-Based Architecture
![Disrupting feedback and integration pathways diminishes sustained neural activity-measured as [latex] NsN^{s}AUC [/latex]-in ipsundrum variants, suggesting a critical role for these mechanisms in maintaining post-stimulus neural persistence.](https://arxiv.org/html/2602.23232v1/2602.23232v1/x4.png)
Researchers have unveiled a small, inspectable AI system designed to explore the core mechanisms underlying persistent thought, stable preferences, and goal-directed behavior.
Mapping the Nucleus: A Lattice QCD Approach to Structure Functions
![The analysis details five distinct Wick contractions-[latex]C_1C_{\hat{1}}, C_2C_{\hat{2}}, S_1S_{\hat{1}}[/latex] among others-that contribute to the four-point function in equation (9) when applied to baryons, with the specific mathematical form of these contractions dependent on the quark flavor of the involved currents.](https://arxiv.org/html/2602.22379v1/2602.22379v1/x1.png)
New calculations using lattice quantum chromodynamics and four-point correlation functions provide a direct path to understanding the internal structure of protons and neutrons.