The rapid advancement of artificial intelligence (AI) technologies has led to an increasing deployment of AI models on edge and terminal devices, driven by the proliferation of the Internet of Things (IoT) and the need for real-time data processing. This survey comprehensively explores the current state, technical challenges, and future trends of on-device AI models. We define on-device AI models as those designed to perform local data processing and inference, emphasizing their characteristics such as real-time performance, resource constraints, and enhanced data privacy. The survey is structured around key themes, including the fundamental concepts of AI models, application scenarios across various domains, and the technical challenges faced in edge environments. We also discuss optimization and implementation strategies, such as data preprocessing, model compression, and hardware acceleration, which are essential for effective deployment. Furthermore, we examine the impact of emerging technologies, including edge computing and foundation models, on the evolution of on-device AI models. By providing a structured overview of the challenges, solutions, and future directions, this survey aims to facilitate further research and application of on-device AI, ultimately contributing to the advancement of intelligent systems in everyday life.

Matrix completion is fundamental for predicting missing data with a wide range of applications in personalized healthcare, e-commerce, recommendation systems, and social network analysis. Traditional matrix completion approaches typically assume centralized data storage, which raises challenges in terms of computational efficiency, scalability, and user privacy. In this paper, we address the problem of federated matrix completion, focusing on scenarios where user-specific data is distributed across multiple clients, and privacy constraints are uncompromising. Federated learning provides a promising framework to address these challenges by enabling collaborative learning across distributed datasets without sharing raw data. We propose \texttt{FedMC-ADMM} for solving federated matrix completion problems, a novel algorithmic framework that combines the Alternating Direction Method of Multipliers with a randomized block-coordinate strategy and alternating proximal gradient steps. Unlike existing federated approaches, \texttt{FedMC-ADMM} effectively handles multi-block nonconvex and nonsmooth optimization problems, allowing efficient computation while preserving user privacy. We analyze the theoretical properties of our algorithm, demonstrating subsequential convergence and establishing a convergence rate of $\mathcal{O}(K^{-1/2})$, leading to a communication complexity of $\mathcal{O}(\epsilon^{-2})$ for reaching an $\epsilon$-stationary point. This work is the first to establish these theoretical guarantees for federated matrix completion in the presence of multi-block variables. To validate our approach, we conduct extensive experiments on real-world datasets, including MovieLens 1M, 10M, and Netflix. The results demonstrate that \texttt{FedMC-ADMM} outperforms existing methods in terms of convergence speed and testing accuracy.

In multi-turn dialogues, large language models (LLM) face a critical challenge of ensuring coherence while adapting to user-specific information. This study introduces the persona knowledge gap, the discrepancy between a model's internal understanding and the knowledge required for coherent, personalized conversations. While prior research has recognized these gaps, computational methods for their identification and resolution remain underexplored. We propose Conversation Preference Elicitation and Recommendation (CPER), a novel framework that dynamically detects and resolves persona knowledge gaps using intrinsic uncertainty quantification and feedback-driven refinement. CPER consists of three key modules: a Contextual Understanding Module for preference extraction, a Dynamic Feedback Module for measuring uncertainty and refining persona alignment, and a Persona-Driven Response Generation module for adapting responses based on accumulated user context. We evaluate CPER on two real-world datasets: CCPE-M for preferential movie recommendations and ESConv for mental health support. Using A/B testing, human evaluators preferred CPER's responses 42% more often than baseline models in CCPE-M and 27% more often in ESConv. A qualitative human evaluation confirms that CPER's responses are preferred for maintaining contextual relevance and coherence, particularly in longer (12+ turn) conversations.

A Florida couple is behind bars for allegedly using an online dating app to lure a 16-year-old girl to their home, brutally torturing and murdering her before dismembering her remains. The body of Miranda Corsette was discarded in a dumpster days after she was reported missing on Feb. 24, according to the St. Petersburg Police Department. Authorities allege that Steven Gress, 35, used the online dating app Grindr to lure Corsette to his house, located approximately 20 miles southwest of Tampa, on Feb. 14. "After meeting him the first time, [Corsette] went home and then the next day she returned to [Gress'] home," police said. Miranda Corsette smiles in an undated photograph shared by the St. Petersburg Police Department. Corsette was allegedly murdered by a man she had met on a dating app, Steven Gress, and his domestic partner, Michelle Brandes.

In recommender systems, the patterns of user behaviors (e.g., purchase, click) may vary greatly in different contexts (e.g., time and location). This is because user behavior is jointly determined by two types of factors: intrinsic factors, which reflect consistent user preference, and extrinsic factors, which reflect external incentives that may vary in different contexts. Differentiating between intrinsic and extrinsic factors helps learn user behaviors better. However, existing studies have only considered differentiating them from a single, pre-defined context (e.g., time or location), ignoring the fact that a user's extrinsic factors may be influenced by the interplay of various contexts at the same time. In this paper, we propose the Intrinsic-Extrinsic Disentangled Recommendation (IEDR) model, a generic framework that differentiates intrinsic from extrinsic factors considering various contexts simultaneously, enabling more accurate differentiation of factors and hence the improvement of recommendation accuracy. IEDR contains a context-invariant contrastive learning component to capture intrinsic factors, and a disentanglement component to extract extrinsic factors under the interplay of various contexts. The two components work together to achieve effective factor learning. Extensive experiments on real-world datasets demonstrate IEDR's effectiveness in learning disentangled factors and significantly improving recommendation accuracy by up to 4% in NDCG.

The writing was already on the wall, but now it's official: The Google Assistant era is over. In a blog post Friday, Google announced plans for Google Assistant's final phase-out, starting on mobile and continuing with tablets, cars, and mobile-connected devices such as headphones and tablets. Finally, Google Assistant will be going away on Nest smart speakers and displays as well as on Google TV devices. Google Assistant's replacement will, of course, be Gemini, Google's entry in the generative AI race. Gemini itself will become the new assistant on Google mobile devices such as phones and tablets, while a "new experience powered by Gemini" is coming to smart speakers and displays.

Google Assistant's days are numbered. Google announced Friday that all Android devices are switching to Gemini as their default assistant and "the classic Google Assistant will no longer be accessible on most mobile devices." The company says it's working to convert more mobile devices from Google Assistant to Gemini in 2025, and plans on "upgrading tablets, cars and devices that connect to your phone, such as headphones and watches" to the new AI assistant. That presumably includes other platforms like iOS, as well. While smart home devices don't seem to be a focus at Google as of late, the company also reaffirmed plans to use Gemini to power a new experience on speakers, displays, and streaming boxes.

Writing about smart home technology, smart devices, and voice assistants is my job. Yet, I don't remember the last time I actually spoke with Alexa. Just to be clear, I don't mean to pick on Alexa per se. I rarely speak to Google Assistant or Apple's Siri, either. It's way easier to haul out my phone and use an app than it is to get a supposedly "smart" voice assistant to do what I want.

In this episode, we dive into Devindra's review of the excellent M4-equipped MacBook Air (and briefly chat about the new Mac Studio). We also discuss Apple's surprise announcement that it's delaying its smarter, AI-infused Siri, which may not arrive until next year. Did Apple over-promise last year, or is it wise to hold off on advanced AI features until they're ready? After all, Apple doesn't want a fiasco like Microsoft's Recall announcement. Listen below or subscribe on your podcast app of choice.

This paper introduces a novel dataset REGEN (Reviews Enhanced with GEnerative Narratives), designed to benchmark the conversational capabilities of recommender Large Language Models (LLMs), addressing the limitations of existing datasets that primarily focus on sequential item prediction. REGEN extends the Amazon Product Reviews dataset by inpainting two key natural language features: (1) user critiques, representing user "steering" queries that lead to the selection of a subsequent item, and (2) narratives, rich textual outputs associated with each recommended item taking into account prior context. The narratives include product endorsements, purchase explanations, and summaries of user preferences. Further, we establish an end-to-end modeling benchmark for the task of conversational recommendation, where models are trained to generate both recommendations and corresponding narratives conditioned on user history (items and critiques). For this joint task, we introduce a modeling framework LUMEN (LLM-based Unified Multi-task Model with Critiques, Recommendations, and Narratives) which uses an LLM as a backbone for critiquing, retrieval and generation. We also evaluate the dataset's quality using standard auto-rating techniques and benchmark it by training both traditional and LLM-based recommender models. Our results demonstrate that incorporating critiques enhances recommendation quality by enabling the recommender to learn language understanding and integrate it with recommendation signals. Furthermore, LLMs trained on our dataset effectively generate both recommendations and contextual narratives, achieving performance comparable to state-of-the-art recommenders and language models.