Papers where the original authors have engaged with our plain-language explanation.
For every paper on this page, at least one of the original authors has seen our plain-language explanation and engaged with it — either confirming it reads accurately or requesting corrections that we then applied. An endorsement does not mean the authors formally approve every sentence, but it does mean the explanation has passed the eyes of the people who wrote the paper.
438 papers reviewed by authors · 251–260 / 438
This paper investigates the gap between industry marketing and user realities of AI agents by analyzing 102 commercial tools and conducting a usability study with 31 participants, revealing that while users are impressed, they face significant challenges due to misaligned capabilities and a lack of meta-cognitive collaboration skills.
This paper establishes a comprehensive mathematical framework for risk-averse ensemble control of control-affine systems by proving the existence of optimal solutions, characterizing the control-to-state mapping with strong regularity properties to derive first-order optimality conditions, and validating the theory through a quantum control application.
By analyzing over 300,000 employer reviews using the Competing Values Framework and CultureBERT, this study demonstrates that while employment verification on platforms like Blind alters the representation of organizational culture compared to anonymous platforms like Glassdoor, it shifts rather than eliminates bias, thereby providing job seekers with systematically different cultural signals that influence their application decisions.
This paper introduces Adaptive Pluralistic Alignment (APA), a modular and efficient pipeline that dynamically updates AI systems to track evolving societal values by learning compact personalized reward models and employing a social-choice-theoretic voting jury, thereby avoiding value lock-in without requiring costly retraining.
This paper proposes a unified framework for Human-AI collaboration that integrates multimodal alignment, interaction-centric explainability, and agency-preserving mechanisms to ensure user trust and control as interfaces evolve from screen-based to multimodal systems.
The paper introduces the Antipodal Method, a novel algorithm that computes generalized winding numbers for 3D surfaces with arbitrary precision and exceptional speed by reformulating the problem as a sum of signed ray intersections and a boundary integral, thereby overcoming the accuracy-efficiency trade-offs of existing approaches for both meshes and parametric surfaces.
This paper challenges the finding of "astonishing regularity" in student learning rates by demonstrating that estimates of learning-rate heterogeneity are highly sensitive to practice-sequence length, with shorter observation windows artificially inflating the perceived variability in student learning speeds.
This paper demonstrates that while incomplete observation of donor actions does not alter the conditions for cooperation in indirect reciprocity, reputation fading hinders it, though the introduction of costly punishment can effectively sustain cooperation and efficiency specifically under reputation fading conditions.
By combining a unified atomistic framework with ultrahigh-field measurements on carbon nanotube fibers, this study reveals that macroscopic transport in disordered low-dimensional networks is primarily governed by junction-level quantum interference, where positive magnetoresistance stems from junction overlap and negative magnetoresistance arises from lattice-mismatched heterojunctions.
This paper investigates phenomenological constraints and future discovery potential for muonphilic portals to fermionic asymmetric dark matter, analyzing both effective field theory operators and specific UV models to determine how 3 and 10 TeV muon colliders can probe parameter spaces currently allowed by direct detection, collider limits, and muon anomalies.