Abstract Large-scale artificial intelligence (AI) models are fundamentally transforming industries and redefining the paradigm of human–machine collaboration. While the technological revolution signals a new era of machine intelligence, the continued scaling of these models has exposed significant limitations in contemporary hardware architectures, manifesting as constraints on computational efficiency, interconnection bandwidth, and memory capacity. These three dimensions are inseparably intertwined, such that advances along any single axis often exacerbate bottlenecks in the others, rendering isolated optimizations increasingly ineffective. Achieving an optimal balance among them to maximize system efficiency therefore remains a central challenge in the design of scalable AI systems. To address this challenge, we introduce Computation-Bandwidth-Memory Trade-offs, termed the AI Trinity, a unified paradigm that positions computation , bandwidth , and memory as coequal pillars for next-generation AI infrastructure. Inspired by the device-edge-cloud collaboration principle from the AI Flow framework, we formulate AI Trinity as a resource-theoretic view of the computation-bandwidth-memory bottlenecks in distributed AI systems. Within this framework, AI Trinity identifies three fundamental trade-offs: (1) More Computation $$\rightarrow$$ → Less Bandwidth, wherein computational resources are exploited to reduce data transmission under limited bandwidth conditions, (2) More Bandwidth $$\rightarrow$$ → Less Memory, which exploits abundant communication capacity to populate or refresh memory when local storage resources are constrained, and (3) More Memory $$\rightarrow$$ → Less Computation, whereby storage capacity are utilized to mitigate redundant computation when computational costs are prohibitive. We illustrate its effectiveness through representative system designs spanning edge–cloud communication, large-scale distributed training, and model inference. The innovations embodied in AI Trinity advance a new paradigm for scalable AI infrastructure, providing both a conceptual foundation and practical guidance for a broad range of application scenarios.
Our students and classrooms have changed significantly in the last few years. Technology has reshaped how we teach and learn. Artificial intelligence is changing how students engage with information. At the same time, higher education classrooms continue to bring together learners with different experiences, expectations and approaches to learning. So, has the way we assess learning kept pace? This session takes a closer look at what assessment can look like in today’s higher education classroom, with a focus on authentic assessment and intentional course design. We’ll consider how assessment can move beyond simply measuring what students remember and instead create opportunities for students to apply what they know in meaningful ways. The session will also explore the opportunities and challenges AI brings to assessment along with practical considerations for designing courses that support a multigenerational classroom and encourage student engagement. Whether you are rethinking an existing assessment or starting from scratch, this session offers ideas and strategies for taking a more intentional approach to assessment in the 21st century classroom.
Ashton Hays· Scholar Works at Harding (Ha...· 0 citations
Artificial Intelligence (AI) is increasingly embedded in organisational life, and this dissertation explores employees’ experience of AI in everyday work. It focuses on broad-application AI, commonly introduced through organisational initiatives led by HR, Learning & Development, or IT with the aim of supporting employees. Rather than approaching AI as a discrete tool with fixed effects, the dissertation examines unfolding relations in specific contexts and ask what matters in AI-inclusive work. This dissertation shows that AI applications are not merely sets of functionalities, but are inseparable from the situated context, shaping and being shaped by employees, work practices, and the organisational environment. It finds that conversational AI, using natural language instead of a menu-based interface for employee self-service, can contribute to a motivation and well-being supportive organisational environment by fostering greater autonomy, competence and relatedness. It also examines how employees interact with AI systems internal and external to their organisations (including unendorsed “shadow AI”), such as contextual search, content recommendations and generative AI in knowledge work. These interactions, shaped by past habits, present constraints and imagined futures, gradually reshape the boundaries of tasks, relationships and the meaning of work. Finally, the dissertation argues that different types of AI matter in different ways because they elicit different forms of engagement and different workplace experiences. Discriminative AI is positioned as a tool for the task, foregrounding efficiency and effectiveness, while also potentially giving rise to possible negative long-term experiences and raising questions about meaningful work. Generative AI is positioned as a medium for creative expression, foregrounding exploration, innovation, and creative actions, thereby fostering more creative and positive experiences at work. Overall, this dissertation offers insights for scholars and practitioners into emerging relations in AI-inclusive work, showing that the value and effects of AI do not reside in technology alone, but emerge through the relations among employees, AI characteristics and organisational context. In doing so, it offers a perspective that moves beyond short-term gains and highlights the longer-term value of creating work environments in which AI is experienced as useful, meaningful and supportive.
Dijana; id_orcid 0009-0005-2046-9468 Aleksić· EUR Research Repository (Era...· 0 citations
The increasing adoption of generative artificial intelligence (AI) in software development introduces a structural asymmetry: implementation generation now occurs at machine speed, while behavioral governance remains constrained by institutional deliberation. These processes are not merely mismatched in velocity; they are orthogonal in function. Generation produces executable artifacts. Governance establishes permissible behavior. This paper formalizes the generation-governance impedance mismatch as a structural property of AI-accelerated systems. We argue that conventional governance mechanisms—code review, testing, and audit processes—operate at the implementation layer and therefore cannot scale to match machine-speed generation without structural reform.
Bhash Ganti, Bhash Ganti· Zenodo (CERN European Organi...· 0 citations
Reach audiences
Advertise in front of researchers, engineers, and readers.
The increasing adoption of generative artificial intelligence (AI) in software development introduces a structural asymmetry: implementation generation now occurs at machine speed, while behavioral governance remains constrained by institutional deliberation. These processes are not merely mismatched in velocity; they are orthogonal in function. Generation produces executable artifacts. Governance establishes permissible behavior. This paper formalizes the generation-governance impedance mismatch as a structural property of AI-accelerated systems. We argue that conventional governance mechanisms—code review, testing, and audit processes—operate at the implementation layer and therefore cannot scale to match machine-speed generation without structural reform.
Bhash Ganti, Bhash Ganti· Zenodo (CERN European Organi...· 0 citations
Autonomous flight in unknown cluttered environments is hindered by the computation-quality-memory trilemma of onboard trajectory generation. In this paper, we propose an efficient end-to-end local planner via imitation learning. A lightweight offline-primitive-based dataset collection framework is designed to produce safe and high-quality trajectory primitives in non-convex environments. A compact neural network directly maps sensory inputs to polynomial coefficients that inherently encode higher-order dynamical information. The learned policy generates smooth, empirically collision-free and dynamically feasible trajectories in real time without back-end solving. It achieves ultra-fast computation (below 1ms on a standard desktop and average 3.68ms during onboard flight), while maintaining low onboard memory requirements (less than 1.5MiB). Extensive simulation benchmarks demonstrate superiority in both planning latency and target-reaching progress quality. Zero-shot deployment in real-world experiments further validates the robust sim-to-real transfer capability of the proposed method.
Zhitao Liu, Guangtong Xu, Zihan Wang et al.· 0 citations
Conditional segmentation models may be trained and evaluated with auxiliary signals cleaner than those available at deployment. We study this protocol-level manifestation of shortcut learning and auxiliary-variable shift in phase-conditioned echocardiographic segmentation. The complementary gap pair measures loss on the deployable oracle-estimated pathway and probes sensitivity on the oracle-random pathway. On held-out CAMUS data, one strong-cyclic, oracle-selected run fails severely with estimated phase, while sensitivity to incorrect phase persists across three runs. On EchoNet-Dynamic, the current estimator remains usable, but random-phase testing reveals strong latent sensitivity. Deployment-aware checkpoint selection and phase perturbation reduce both gaps with little change in mean Dice. Exploratory subgroup analyses quantify variation across measured strata, and a downstream ejection fraction (EF) audit shows that recovering segmentation does not necessarily recover EF error or signed bias. Together, the gaps test whether oracle-conditioned performance survives the inference pathway actually available at deployment.
Dang P. M. Cao, Hieu D. Pham, Hieu Pham· 0 citations
Parkinson's disease (PD) selectively impairs distinct stages of motor control. Using backspace events as natural error-correction episodes in the public neuroQWERTY MIT-CSXPD dataset (n=57 subjects, 27 PD with UPDRS-III scores), we test whether passively-collected keystroke timing dissociates a variability-based pre-error monitoring signal from a speed-based post-error recovery signal. Pre-error typing instability does not track PD severity (r=-0.072, p=0.721), while post-error pause duration does (r=+0.656, p=0.0002; subject-level OLS p=1.2x10^-4, n=27, primary analysis given within-subject event clustering). A mixed-effects log-normal model with a random subject intercept confirms this while retaining full event-level power (coef=0.0250, p=1.2x10^-4, n=1,563 events), closely matching the subject-level OLS despite an unrelated estimation strategy. Error-detection latency also correlates with UPDRS-III (r=+0.660), confirming the dissociation is between variability and speed, not detection and correction per se. A log-normal accelerated failure time model yields a coefficient of 0.0255 (bootstrap 95% CI [0.0146, 0.0357]), a 2.6% increase per UPDRS-III point. After controlling the immediately thecoefficient attenuates to 0.0133 (p=3.1x10^-21),consistent w timingincrement above general bradykinesia. Results reacross both S2:r=+0.645), survive jackknife exclusion of every subject, andnatingfinger-tapping and mPower smartphone tapping (group AUC=0.836). cellent(ICC(2,1)=0.945, n=20).
We present Anamnesis, an interactive system for demographically controllable survey simulation using large language models. Open-source and designed for non-technical users/researchers, Anamnesis enables the prototyping and stress-testing of survey instruments on virtual populations rather than real human subjects. The platform operationalizes the recently introduced Anthology and Alterity frameworks, which use structured narrative backstories to condition model responses, within a unified web interface. It supports open-ended generation, probabilistic demographic resampling, and multimodal (image and audio) surveys. We evaluate the system through two case studies: (1) replicating segments of Pew Research Center's American Trends Panel (ATP) on political typology and biomedical issues and (2) emulating human preference in the New Yorker Caption Contest. In both cases, Anamnesis produces opinion distributions that more closely match real-world survey data than standard persona-prompting baselines, offering a transparent, reproducible, and open-source alternative to proprietary simulation services.
Song-Ze Yu, Joseph Suh, Serina Chang et al.· 0 citations
Creating photorealistic, animatable 3D human avatars from monocular images still largely depends on Linear Blend Skinning (LBS) and parametric body models, which constrain expressivity and often introduce artifacts due to imperfect fitting. We propose LUNA, an LBS-free universal neural animation model that directly maps multiple 2D controls like images, keypoints, sketches, and unseen characters into 3D Gaussian deformations, bypassing explicit body fitting. At its core, a transformer-based motion regressor disentangles global rigid motion from fine-grained local dynamics to capture both coherent movement and subtle non-rigid effects. To resolve the inherent ambiguity of 2D-to-3D lifting while scaling beyond fitted datasets, we introduce hybrid supervision that distills soft structural priors from an LBS teacher and a loss that supports training on both limited fitted data and large in-the-wild unlabeled videos. Extensive experiments show LUNA achieves competitive visual fidelity compared to LBS-based approaches, while delivering realistic human motion and zero-shot cross-identity generalization across diverse driving modalities. To the best of our knowledge, LUNA is the first end-to-end 3D animatable model that supports implicit 2D driving.
Peng Li, Rawal Khirodkar, Junxuan Li et al.· 0 citations
Agents are increasingly capable of automating software tasks, but can they teach humans how to use software themselves? We introduce DigitalCoach, a multimodal dataset of 72 human expert-novice computer use coaching sessions consisting of 22,752 dialogue turns grounded in 28.1 hours of screen and input event recordings across five software applications. We use DigitalCoach to evaluate whether state-of-the-art models can teach humans how to use computers. Automated evaluation shows that models differ from humans in how they coach: models provide more direct instructions, but fewer explanations, error diagnoses, and knowledge-check questions. When we fix the coaching method, models produce utterances similar to human references yet poorly grounded in visual context. Interactive evaluation confirms that model coaches cause learners to passively follow instructions without deeper engagement and fall short in visual grounding. DigitalCoach lays a foundation for collaborative and proactive computer use coaching agents. Data and code are available at https://project-digital-coach.vercel.app.
Meng Chen, Anya Ji, Tsung-Han Wu et al.· 0 citations
As LLMs increasingly serve as moral advisors and agents, they must address conflicts between competing values. Yet prior work on moral dilemmas overlooks a central aspect of human moral cognition: imagining alternatives beyond the given options. We introduce MoralAltDataset, comprising 307 Advisor and AI-facing Agent dilemmas augmented with compromise and reframed alternatives. We compare human and LLM judgments in binary and four-option settings. Across human participants and 15 LLMs, aggregate moral choice distributions differ substantially between the two settings, with compromise often preferred over either original binary option. Results show value shifts and stronger human-LLM agreement on alternatives. Source-stratified results reveal a descriptive gap: human alternative-selection rates are similar across authoring sources, whereas LLMs select GPT-5-authored alternatives substantially more often. We then compare human-authored alternatives with outputs from three representative LLMs through pairwise preference and expert-based evaluations. Alternatives from these LLMs are generally preferred and better satisfy fine-grained structural and ethical criteria, while revealing a trade-off between structural quality and practical feasibility. Our dataset is available here: https://huggingface.co/datasets/jongchanch/MoralAltDataset, and our project page is here: https://jongchanchoi.com/moral-imagination
Jongchan Choi, Nari Yang, Sung Soo Park et al.· 0 citations
