Explore the vast range of titles available.

This paper reimagines AI hallucinations, instances where large language models (LLMs) generate coherent yet factually ungrounded content, as posthuman hermeneutic practices, where human and machine agencies entangle to produce new modes of meaning-making. Moving beyond binary framings of hallucinations as flaws or features, the analysis situates them within Donna Haraway's sympoietic ethics and proposes sympoietic creativity between human and AI, arguing that these generative anomalies are not errors but provocations for rethinking creativity as a distributed and collaborative act. Through case studies of models like DeepSeek-R1 and artistic projects such as Pharmako-AI (2021), this article demonstrates how hallucinations act as hermeneutic knots, sites where algorithmic noise is curated into cultural critique and interspecies narratives. By contrasting human creativity that is rooted in intentionality and embodiment with AI's stochastic recombination, this paper exposes the ontological chasm between anthropocentric artistry and machine-generated “spectral pantomimes.” Ultimately, the article challenges stakeholders to embrace AI's disruptive potential, not as a tool for replication but as a mediator for posthuman futures, where creativity is redefined through justice, relationality, and the radical interdependence of human and non-human agents.

Formation of the autophagosome requires significant membrane input from cellular organelles. However, no direct evidence has been developed to link autophagic factors and the mobilization of membranes to generate the phagophore. Previously, we established a cell-free LC3 lipidation reaction to identify the ER-Golgi intermediate compartment (ERGIC) as a membrane source for LC3 lipidation, a key step of autophagosome biogenesis (Ge et al., eLife 2013; 2:e00947). We now report that starvation activation of autophagic phosphotidylinositol-3 kinase (PI3K) induces the generation of small vesicles active in LC3 lipidation. Subcellular fractionation studies identified the ERGIC as the donor membrane in the generation of small lipidation-active vesicles. COPII proteins are recruited to the ERGIC membrane in starved cells, dependent on active PI3K. We conclude that starvation activates the autophagic PI3K, which in turn induces the recruitment of COPII to the ERGIC to bud LC3 lipidation-active vesicles as one potential membrane source of the autophagosome.

Autophagy is a catabolic process for bulk degradation of cytosolic materials mediated by double-membraned autophagosomes. The membrane determinant to initiate the formation of autophagosomes remains elusive. Here, we establish a cell-free assay based on LC3 lipidation to define the organelle membrane supporting early autophagosome formation. In vitro LC3 lipidation requires energy and is subject to regulation by the pathways modulating autophagy in vivo. We developed a systematic membrane isolation scheme to identify the endoplasmic reticulum–Golgi intermediate compartment (ERGIC) as a primary membrane source both necessary and sufficient to trigger LC3 lipidation in vitro. Functional studies demonstrate that the ERGIC is required for autophagosome biogenesis in vivo. Moreover, we find that the ERGIC acts by recruiting the early autophagosome marker ATG14, a critical step for the generation of preautophagosomal membranes. Cells continually adapt their behavior to accommodate changes in their environment. For example, when nutrients are abundant, cells can grow or proliferate; in times of scarcity, however, they must conserve resources for essential tasks. In particular, during periods of starvation, cells can cannibalize themselves in a process called autophagy, which literally means ‘self-eating’. Structures called autophagosomes engulf bits of cytoplasm and carry the contents to the digestive compartment of the cell, the lysosome, to be broken down into their constituent parts. This can include the degradation of proteins into amino acids, which can then be recycled into other proteins needed by the cell. In cells, proteins are shipped to their destinations—which can be the plasma membrane or a specific organelle within the cell—via a delivery system known as the secretory pathway. This pathway begins in the endoplasmic reticulum (ER), where many of these proteins are made. From the ER, the proteins move to a compartment called the Golgi apparatus, which then sends them to their destinations, or to the lysosome to be broken down. Between the ER and Golgi they pass through a structure called the ER–Golgi intermediate compartment (ERGIC). Although the signaling pathways that initiate autophagy are known, less is understood about the actual formation of the autophagosomes. Now, Ge et al. have developed an in vitro system to study their formation, and gone on to identify a membrane that is both necessary and sufficient to create these structures. Previous studies have implicated a variety of membranes—including the plasma membrane and the membranes belonging to the ER, the Golgi apparatus, the lysosome and various other organelles—in the formation of autophagosomes. To identify which of these membranes might be involved, Ge et al. focused on a protein called LC3 that is a key marker for the formation of the autophagosome. This protein is recruited to the growing autophagosome by a lipid, so discovering which membranes can add a lipid to LC3 should shed light on the assembly process. By separating the full range of organelles in a cell lysate into fractions (a process called fractionation), Ge et al. found that the ERGIC was the most active membrane to attach lipid to LC3. Additionally, the lipid was only added when signaling pathways that stimulate autophagy—such as the PI3K pathway—were activated. Together, these results provide insight into the mechanism of autophagosome formation, and the structures in the cell that participate in this process.

HighlightsUltrathin (< 600 nm) and defect-free leaf-like UiO-66-SO3H membranes were fabricated via in situ smart growth.The sulfonated angstrom-sized ion transport channels in the membranes could accelerate the cation permeation (~ 3×  faster than non-functionalized UiO-66 membrane) and achieve a high ion selectivity (Na+/Mg2+ > 140).Metal–organic frameworks (MOFs) with angstrom-sized pores are promising functional nanomaterials for the fabrication of cation permselective membranes (MOF-CPMs). However, only a few research reports show successful preparation of the MOF-CPMs with good cation separation performance due to several inherent problems in MOFs, such as arduous self-assembly, poor water resistance, and tedious fabrication strategies. Besides, low cation permeation flux due to the absence of the cation permeation assisting functionalities in MOFs is another big issue, which limits their widespread use in membrane technology. Therefore, it is necessary to fabricate functional MOF-CPMs using simplistic strategies to improve cation permeation. In this context, we report a facile in situ smart growth strategy to successfully produce ultrathin (< 600 nm) and leaf-like UiO-66-SO3H membranes at the surface of anodic alumina oxide. The physicochemical characterizations confirm that sulfonated angstrom-sized ion transport channels exist in the as-prepared UiO-66-SO3H membranes, which accelerate the cation permeation (~ 3× faster than non-functionalized UiO-66 membrane) and achieve a high ion selectivity (Na+/Mg2+ > 140). The outstanding cation separation performance validates the importance of introducing sulfonic acid groups in MOF-CPMs.

Unlike inorganic crystals, metal-organic frameworks do not have a well-developed nanostructure library, and establishing their appropriately diverse and complex architectures remains a major challenge. Here, we demonstrate a general route to control metal-organic framework structure by a solvent-assisted ligand exchange approach. Thirteen different types of metal-organic framework structures have been prepared successfully. To demonstrate a proof of concept application, we used the obtained metal-organic framework materials as precursors for synthesizing nanoporous carbons and investigated their electrochemical Na + storage properties. Due to the unique architecture, the one-dimensional nanoporous carbon derived from double-shelled ZnCo bimetallic zeolitic imidazolate framework nanotubes exhibits high specific capacity as well as superior rate capability and cycling stability. Our study offers an avenue for the controllable preparation of well-designed meta-organic framework structures and their derivatives, which would further broaden the application opportunities of metal-organic framework materials. Metal-organic frameworks are promising for a range of applications, but architectural control is challenging. Here the authors use solvent-assisted ligand exchange to access a variety of metal-organic framework nanomaterials for precursors of nanoporous carbon with sodium ion storage properties.

Recent evidence suggests that autophagy facilitates the unconventional secretion of the pro-inflammatory cytokine interleukin 1β (IL-1β). Here, we reconstituted an autophagy-regulated secretion of mature IL-1β (m-IL-1β) in non-macrophage cells. We found that cytoplasmic IL-1β associates with the autophagosome and m-IL-1β enters into the lumen of a vesicle intermediate but not into the cytoplasmic interior formed by engulfment of the autophagic membrane. In advance of secretion, m-IL-1β appears to be translocated across a membrane in an event that may require m-IL-1β to be unfolded or remain conformationally flexible and is dependent on two KFERQ-like motifs essential for the association of IL-1β with HSP90. A vesicle, possibly a precursor of the phagophore, contains translocated m-IL-1β and later turns into an autophagosome in which m-IL-1β resides within the intermembrane space of the double-membrane structure. Completion of IL-1β secretion requires Golgi reassembly and stacking proteins (GRASPs) and multi-vesicular body (MVB) formation. Cells release a large number of proteins to the extracellular space. The majority of these ‘secreted’ proteins first pass through two structures inside cells called the endoplasmic reticulum and Golgi. However, a growing number of proteins have been identified that are released by an unconventional mechanism that bypasses the endoplasmic reticulum and Golgi. Autophagy is a process that destroys damaged proteins and other unwanted material in cells. It gets triggered when cells are starved of nutrients, leading them to digest their own materials and recycle the resources into new molecules. During autophagy, a cup-like structure with a double layer of membrane forms around the material that is to be digested. This structure then elongates and eventually engulfs the material to form a bubble-like compartment called the autophagosome. Recent evidence has suggested that autophagosomes are involved in the unconventional secretion of a protein called interleukin-1β; this protein is crucial for the body’s immune response against infection. However, it was not clear how these proteins entered the autophagosomes. Zhang et al. have now explored the link between interleukin-1β and autophagy in more detail. The experiments showed that when autophagy was triggered by starvation, the secretion of interleukin-1β was enhanced. Conversely, when autophagy was inhibited, interleukin-1β accumulated inside the cells and could not be secreted. Further experiments then revealed unexpectedly that interleukin-1β was not engulfed by the cup-like structure (as is the case for material that is destined to be removed). Instead, interleukin-1β was found to enter into smaller bubble-like packages (called vesicles) that turn into the autophagosome. Zhang et al. also found that a protein called HSP90 binds to interleukin-1β and enables it to cross the membrane (or translocate) into the vesicles, and that this means that interleukin-1β actually resides in the space between the outer and inner membranes of the autophagosome. How many other proteins share this unusual route out of the cell and what membrane channel is used for this translocation event remain open questions for the future.

Objective This study used machine learning algorithms to identify critical variables and predict postoperative delirium (POD) in patients with degenerative spinal disease. Methods We included 663 patients who underwent surgery for degenerative spinal disease and received general anesthesia. The LASSO method was used to screen essential features associated with POD. Clinical characteristics, preoperative laboratory parameters, and intraoperative variables were reviewed and were used to construct nine machine learning models including a training set and validation set (80% of participants), and were then evaluated in the rest of the study sample (20% of participants). The area under the receiver‐operating characteristic curve (AUROC) and Brier scores were used to compare the prediction performances of different models. The eXtreme Gradient Boosting algorithms (XGBOOST) model was used to predict POD. The SHapley Additive exPlanations (SHAP) package was used to interpret the XGBOOST model. Data of 49 patients were prospectively collected for model validation. Results The XGBOOST model outperformed the other classifier models in the training set (area under the curve [AUC]: 92.8%, 95% confidence interval [CI]: 90.7%–95.0%), validation set (AUC: 87.0%, 95% CI: 80.7%–93.3%). This model also achieved the lowest Brier Score. Twelve vital variables, including age, serum albumin, the admission‐to‐surgery time interval, C‐reactive protein level, hypertension, intraoperative blood loss, intraoperative minimum blood pressure, cardiovascular‐cerebrovascular disease, smoking, alcohol consumption, pulmonary disease, and admission‐intraoperative maximum blood pressure difference, were selected. The XGBOOST model performed well in the prospective cohort (accuracy: 85.71%). Conclusion A machine learning model and a web predictor for delirium after surgery for the degenerative spinal disease were successfully developed to demonstrate the extent of POD risk during the perioperative period, which could guide appropriate preventive measures for high‐risk patients. A machine learning model and a web predictor for delirium after surgery for degenerative spinal disease are successfully developed to demonstrate the extent of POD risk during the perioperative period, which could guide appropriate preventive measures for high‐risk patients.

Air pollution is a serious environmental problem that has attracted much attention. Air quality prediction can provide useful information for urban environmental governance decision-making and residents’ daily health control. However, existing research methods have suffered from a weak ability to capture the spatial correlations and fail to model the long-term temporal dependencies of air quality. To overcome these limitations, we propose a multi-scale spatiotemporal graph convolution network (MST-GCN), which consists of a multi-scale block, several spatial-temporal blocks and a fusion block. We first divide the extracted features into several groups based on their domain categories, and represent the spatial correlations across stations as two graphs. Then we combine the grouped features and the constructed graphs in pairs to form a multi-scale block that feeds into spatial-temporal blocks. Each spatial-temporal block contains a graph convolution layer and a temporal convolution layer, which can model the spatial correlations and long-term temporal dependencies. To capture the group interactions, we use a fusion block to fuse multiple groups. Extensive experiments on a real-world dataset demonstrate that our model achieves the highest performance compared with state-of-the-art and baseline models for air quality prediction.

Understanding active phases across interfaces, interphases, and even within the bulk under varying external conditions and environmental species is critical for advancing heterogeneous catalysis. Describing these phases through computational models faces the challenges in the generation and calculation of a vast array of atomic configurations. Here, we present a framework for the automatic and efficient exploration of active phases. This approach utilizes a topology-based algorithm leveraging persistent homology to systematically sample configurations across diverse coordination environments and material morphologies. Simultaneously, efficient machine learning force fields enable rapid computations. We demonstrate the effectiveness of this framework in two systems: hydrogen absorption in Pd, where hydrogen penetrates subsurface layers and the bulk, inducing a “hex” reconstruction critical for CO 2 electroreduction, explored through 50,000 sampled configurations; and the oxidation dynamics of Pt clusters, where oxygen incorporation renders the clusters less active during oxygen reduction reactions, investigated through 100,000 sampled configurations. In both cases, the predicted active phases and their impacts on catalytic mechanisms closely align with previous experimental observations, indicating that the proposed strategy can model complex catalytic systems and discovery active phases under specific environmental conditions. Discovering active phases in heterocatalysis entails efficient configuration sampling and optimization. Here, the authors developed a framework based on topology and machine learning to effectively explore the active structures, applied in the CO2 electroreduction and Oxygen Reduction Reaction

Injecting carbon dioxide is the most effective means of preventing and extinguishing fires in sealing hazardous areas, but the traditional method slowly and remotely injects carbon dioxide gas into the well after gasification on the ground, which is dependent on the complete mine pipe network without cooling effect. To inject liquid directly from the tank with vacuum interlayer and heat insulating powder for rapid inerting and cooling, a new approach using track mobile platform to go deep into the underground mine disaster area is proposed, so the liquid can be delivered to the nozzle at the end of DN40 large diameter pipe, and the continuous gasification jet can be realized. The experimental results show that: (1) The liquid volume in a tank of vacuum degree within 2.0 Pa and 200 mm interlayer reduced no more than 15.5% after 48 days; (2) Taking the pressure in the tank as the power source, because of environmental differences inside and outside the pipe after 100 m pressure holding delivery, the physical form of liquid and gas could be converted instantly; (3) The continuous discharge time without ice blocking for a tank full of 2 m 3 liquid was about 10.5 min under 25 L dual mode nitrogen pressurization, which is 1/12 of injection time after ground gasification; (4) Based on the temperature decrease trend measured at different positions, the cooling characteristics on liquid gasification jet path are quantified, and the calculation formula of temperature changing with time on the center line of liquid gasification jet is obtained. Through this new approach, the integration of vacuum insulated storage, safe mobile transportation, and continuous and rapid release with large flow can be achieved for the liquid carbon dioxide.