BACKGROUND:Artificial intelligence (AI) tools offer new opportunities to support human perception and provide gaze guidance in surgery. However, there remains a need to better understand surgeons' perceptions of these tools and to identify desired features across levels of expertise. This study explored surgeons' perceptions of AI-based gaze guidance to determine how such tools can best meet surgeons' needs while recognizing their limitations. METHODS:Seventy-one surgeons (38 attendings, 33 residents) watched laparoscopic cholecystectomy videos and responded to open-ended questions about their perceptions of AI for gaze guidance in surgery. Responses were analyzed using latent content analysis to identify themes and desired AI features. Each response was also categorized as positive, negative, or neutral to assess overall perception. RESULTS:Attendings (92%) and residents (82%) were positive about AI for visual gaze guidance. Neutral responses (9.8%) reflected uncertainty or limited familiarity with AI. Desired features included directing attention, identification, highlighting and visualization, feedback, decision support, educational tools, communication and collaboration, and planning and mapping. Identification involved AI support for recognizing critical structures, abnormalities, gaps in attention, and spatial orientation. Attendings favored AI for decision support and feedback, while residents emphasized its value for visualization and education. Surgeons frequently described AI as a "second pair of eyes," providing guidance, visualization, user control, and safety. Design considerations were also identified to provide guidance without being a potential distractor to the surgeon. CONCLUSION:Surgeons across expertise levels expressed positive views toward AI support for gaze guidance, emphasizing the need for features that complement human expertise, align with individual training needs, and account for contextual factors such as minimizing distractions and preventing tunnel vision. Identification, direct attention, and decision support were the most desired features. The findings suggest that tailored AI-based tools can play an important role in surgical training and active procedures by supporting harm reduction.
The deployment of decision-making AI agents presents a critical challenge in maintaining alignment with human values or guidelines while operating in complex, dynamic environments. Agents trained solely to achieve their objectives may adopt harmful behavior, exposing a key trade-off between maximizing the reward function and maintaining alignment. For pre-trained agents, ensuring alignment is particularly challenging, as retraining can be a costly and slow process. This is further complicated by the diverse and potentially conflicting attributes representing the ethical values for alignment. To address these challenges, we propose a test-time alignment technique based on model-guided policy shaping. Our method allows precise control over individual behavioral attributes, generalizes across diverse reinforcement learning (RL) environments, and facilitates a principled trade-off between ethical alignment and reward maximization without requiring agent retraining. We evaluate our approach using the MACHIAVELLI benchmark, which comprises 134 text-based game environments and thousands of annotated scenarios involving ethical decisions. The RL agents are first trained to maximize the reward in their respective games. At test time, we apply policy shaping via scenario-action attribute classifiers to ensure decision alignment with ethical attributes. We compare our approach against prior training-time methods and general-purpose agents, as well as study several types of ethical violations and power-seeking behavior. Our results demonstrate that test-time policy shaping provides an effective and scalable solution for mitigating unethical behavior across diverse environments and alignment attributes.
Existing open-vocabulary detectors focus on RGB images and fail to generalize to thermal imagery, where low texture and emissivity variations challenge RGB-based semantics. We present Thermal-Det, the first large language model (LLM) supervised open-vocabulary detector tailored for thermal images. To enable large-scale training, we construct a synthetic dataset by converting GroundingCap-1M into the thermal domain and filtering captions to remove RGB-specific terms, yielding over one million thermally aligned samples with bounding boxes, grounding texts, and detailed captions. Thermal-Det jointly optimizes detection, captioning, and cross-modal distillation objectives. A frozen RGB teacher provides geometric and semantic pseudo-supervision for paired but unlabeled RGB–thermal data, transferring open-vocabulary knowledge without manual annotation. The model further employs a Thermal–Text Alignment Head for text calibration and a Modality-Fused Cross-Attention module for dual-modality reasoning. Unlike prior domain-adaptation methods, the detector is fully fine-tuned to internalize thermal contrast patterns while preserving language alignment. Experiments on public benchmarks show consistent 2–4\% AP gains over existing open-vocabulary detectors, establishing a strong foundation for scalable, language-driven thermal perception.
Underwater Acoustic Target Recognition (UATR) of ships is well-suited for machine learning, yet its progress is hindered by the lack of large, diverse, and publicly available labeled datasets. In this work, we introduce UniqueShip, a machine learning-ready benchmark dataset for UATR applications sourced from the open Ocean Networks Canada (ONC) repository. Unlike previous datasets, we explicitly control for "data leakage" between the training and evaluation sets to ensure more reliable and generalizable model evaluation that does not encourage the model to memorize individual ships. We demonstrate that typical, random data partitioning in two prominent UATR datasets leads to falsely optimistic test performance, increasing accuracy by 10-48 percentage points compared to our more careful partitioning. Ablations on UniqueShip further show that doubling the number of unique vessels improves accuracy by 2.4-2.6 percentage points, while doubling total audio duration improves only by 0.8-1.3 points, indicating that vessel diversity should drive dataset curation more than total hours. We provide baselines with convolutional and transformer backbones, and analyze how ship metadata correlates with classification performance, finding that individual vessel characteristics predict classification difficulty far better than distance to the hydrophone alone. Overall, UniqueShip contains 2,460 hours of ship-radiated audio from 4,218 unique vessels (3,437 hours including background). We publish the dataset, code, and easy-to-download splits at uniqueshipdata.org to foster further UATR research.
This article describes the application of HEXRD, an open-source Python library for X-ray diffraction analysis, and HEXRDGUI, its companion graphical application, to dynamic compression experiments. Built on a generic multi-detector geometric model, the toolkit supports arbitrary instrument configurations with independently positioned planar or cylindrical detector panels, arbitrary beam directions, and multiple X-ray sources. It calibrates instruments from powder rings, Laue spots or the rings of an unknown phase, fits whole patterns by Le Bail or Rietveld refinement, and imports data from the TARDIS, FIDDLE and PXRDIP platforms at the National Ignition Facility (NIF) and OMEGA EP. Although originally developed for high-energy diffraction microscopy at synchrotron sources, HEXRD has been adopted as an analysis tool for dynamic compression X-ray diffraction experiments at facilities including the NIF, the Matter in Extreme Conditions end-station at the Linac Coherent Light Source, the Dynamic Compression Sector (DCS) at the Advanced Photon Source, OMEGA EP at the Laboratory for Laser Energetics, and the High Energy Density (HED) instrument at the European XFEL. Calibrated from the same ring positions on a single planar detector, HEXRD and the independent code pyFAI predict the same scattering angle 2θ to numerical precision. On a dual-source dataset from the NIF, calibrating the detectors against both X-ray sources at once fits the rings of both to an RMS of 0.18° in 2θ, about one image-plate pixel, whereas a fit to either source alone, with the plate tilts held at their nominal values, fits its own rings nearly as well but mispredicts the other's by about 0.6°. The software is freely available under a BSD license.