哈佛大学

The increasing demand for transparent and reliable models, particularly in high-stakes decision-making areas such as medical image analysis, has led to the emergence of eXplainable Artificial Intelligence (XAI). Post-hoc XAI techniques, which aim to explain black-box models after training, have raised concerns about their fidelity to model predictions. In contrast, Self-eXplainable AI (S-XAI) offers a compelling alternative by incorporating explainability directly into the training process of deep learning models. This approach allows models to generate inherent explanations that are closely aligned with their internal decision-making processes, enhancing transparency and supporting the trustworthiness, robustness, and accountability of AI systems in real-world medical applications. To facilitate the development of S-XAI methods for medical image analysis, this survey presents a comprehensive review across various image modalities and clinical applications. It covers more than 200 papers from three novel perspectives: 1) input explainability through the integration of explainable feature engineering and knowledge graph, 2) model explainability via attention-based learning, concept-based learning, and prototype-based learning, and 3) output explainability by providing textual explanations and visual counterfactual explanations. This paper also outlines desirable characteristics of S-XAI models for assessing explanation quality, while discussing major challenges and future research directions in developing S-XAI for medical image analysis.

High renewable penetration increases the frequency and magnitude of net-load ramps, stressing real-time flexibility. Two commonly deployed remedies are look-ahead economic dispatch (LAED) and ramp products (RPs), yet their operational equivalence under the industry-standard rolling-window dispatch implementation is not well understood. This paper develops linear optimization models for multi-interval LAED and RP-based co-optimization, and proves that an enhanced RP formulation can match LAED's dispatch feasible region at a single time step when additional intertemporal deliverability constraints are enforced. We then show that this equivalence does not generally persist under rolling-window operation because LAED and RP formulations optimize different intertemporal objectives, leading to divergent end-of-window states. Using different test systems under stressed ramping conditions and multiple load levels, we show LAED achieves similar or lower load shedding than RP implementations with the same look-ahead horizon, with the most pronounced differences under high-load, ramp-limited conditions. The study highlights the limitations of current ramp product implementations and suggests enhancements, such as introducing more mid-duration RPs.

Voice phishing (vishing) attacks have traditionally been limited by the need for human operators. The rapid emergence of high-quality AI voice synthesis and large language models (LLMs) reduces this bottleneck and enables scalable, automated scams. In this paper, we conduct a large-scale survey experiment (N=4100) and qualitative interviews (N=12) to assess U.S. adults’ susceptibility to AI-powered voice phishing attacks. Participants were exposed to audio recordings or transcripts of scam scenarios generated using leading voice models such as Llama Full Duplex (Llama FD), Sesame, Gemini, OAI AVM, Play.AI, and ElevenLabs and the corresponding human baselines. The results show high compliance rates. Up to 36% of participants would or might comply with phishing requests in the “relative-in-distress” category. Overall compliance rate across all five scam categories was 16.5%, a striking figure given the low cost and high scalability of AI-automated voice phishing. Caller persuasiveness was the strongest predictor of compliance and certain models (most notably Sesame) achieved ratings comparable to human voices, or sometimes even slightly surpassing them. Our economic analysis suggests that while human-operated vishing is unprofitable at US wages, AI-powered vishing appears to be economically viable for several models. The primary risk of present-day AI-enabled vishing thus lies in the economics of automation rather than novel or “superhuman” persuasive techniques, though these cannot be ruled out for future systems. This raises significant concerns for the design of AI systems, consumer protection, and model release policies.

Kirkland and Šmigoc constructed a family of stochastic matrices realising the Type III boundary polynomials of the Karpelevič region and conjectured that, conversely, every stochastic realisation of such a polynomial must come from their construction. We prove this conjecture for the full nonzero parameter range 0<α≤1, for the reduced Ito polynomialsfα(x)=xy(xq−(1−α))d−αd,n=qd+y, associated with Type III Farey-neighbour arcs of order n. For 0<α<1, the boundary theorem of Dmitriev and Dynkin restricts the possible nonzero entries of a realising matrix. After relabelling the vertices, its directed graph consists of the edges i→i+n1 and a selection of edges i→i+n(n+1−q), with indices taken modulo n. Coates' coefficient formula and a weighted form of Turán's theorem then show that the directed cycles of length q fall into d classes. Any two cycles in the same class share a vertex, while cycles in different classes have no vertices in common. Each class has the same sum of cycle weights, where the weight of a cycle is the product of its edge weights. The positions of these cycles along the n-cycle then yield the product condition required by Kirkland and Šmigoc. The case α=1 is treated separately, and an explanation is provided for the failure when α=0.

As passenger vehicle fleets turn over, lifecycle emissions shift from road transport to the energy and manufacturing sectors, posing new environmental and resource challenges. We propose a fleet-level framework, incorporating dynamic material flow analysis with lifecycle assessment to capture regional and temporal dynamics behind electrification in China. Relative to the baseline scenario, early circular economy implementation could reduce ∼ 50% lifecycle CO2 emissions in 2030. By 2060, integrated strategies could reduce annual fleet lifecycle CO2 emissions by 33%–38%, while recycled lithium could meet 89%–100% of battery demand, complementing the 349.9 Mt CO2 annual reduction achieved by energy-related actions. Substantial regional variations reflect differences in electricity mix, fleet turnover, and local conditions, with Northeast China benefiting more from energy-related actions due to coal-based electricity, whereas the Southwest favors circular economy strategies. Overall, while energy-related actions are effective long-term countermeasures, ambitious circular economy strategies are critical to advancing environmental and resource synergies.