Ai Magazine最新文献

According to an industry survey, many people miss opportunities to apply for government subsidy programs because they do not know how to apply. People also need to search manually and check whether these programs are suitable for them. To address this issue, our study developed a new generative recommender system with both users' information and government subsidy documents. Within our recommender system framework, we modify the existing Residual Quantization Variational Auto-Encoder (RQ-VAE) model to capture deep and abstract information from subsidy documents. Using semantic IDs generated for approximately 185,610 user click-stream histories and 240,000 documents, we train our recommender system to predict the semantic IDs of the next subsidy policy documents in which a user might be interested. In 2024, we successfully deployed our generative recommender system in Wello, a Korean Gov-Tech startup. In collaboration with the Korean government, our generative recommender system helped enhance program effectiveness by saving $7.8 million in unused funds and achieved $27.4 million in advertising efficiency gains. Also, Wello observed a 68% improvement in Click-Through-Ratio (CTR), increasing from 41.4% in the third quarter of 2024 to 69.6% in the fourth quarter of 2024. We thus anticipate that our generative recommender system will have a significant impact on both individuals and the government.

Enterprise AI Assistants are increasingly deployed in domains where accuracy is paramount, making each erroneous output a potentially significant incident. This paper presents a comprehensive framework for monitoring, benchmarking, and continuously improving such complex, multi-component systems under active development by multiple teams. Our approach encompasses three key elements: (1) a hierarchical “severity” framework for incident detection that identifies and categorizes errors while attributing component-specific error rates, facilitating targeted improvements; (2) a scalable and principled methodology for benchmark construction, evaluation, and deployment, designed to accommodate multiple development teams, mitigate overfitting risks, and assess the downstream impact of system modifications; and (3) a continual improvement strategy leveraging multidimensional evaluation, enabling the identification and implementation of diverse enhancement opportunities. By adopting this holistic framework, organizations can systematically enhance the reliability and performance of their AI Assistants, ensuring their efficacy in critical enterprise environments. We conclude by discussing how this multifaceted approach opens avenues for various classes of enhancements, including human-AI collaborative evaluation, paving the way for more robust and trustworthy AI systems.

This year's innovative applications of AI special issue features AI systems deployed in real-world settings, from enterprise platforms to public services, demonstrating both technical rigor and measurable benefits for organizations and society. The eight selected articles span enterprise reliability, cybersecurity, aerospace, education, healthcare logistics, government services, and scalable AI strategy. Collectively, these works illustrate how AI is progressing from research prototypes to systems that organizations now rely on for critical decisions, offering lessons learned for both researchers and practitioners.

Finetuning serves as the critical adaptation mechanism for multimodal large language models, bridging their pretrained knowledge with specialized downstream task requirements. This paper reviews recent finetuning advances across three key dimensions: (1) efficiency-oriented methods that reduce resource costs; (2) capability-specific techniques enhancing specialized multimodal skills; and (3) task-unifying approaches that bridge understanding and generation. We demonstrate how these directions transform multimodal large language models from versatile foundations into adaptive, human-aligned systems, providing researchers with a structured roadmap for developing next-generation multimodal AI.

Ensuring that AI systems do what we, as humans, actually want them to do is one of the biggest open research challenges in AI alignment and safety. My research seeks to directly address this challenge by enabling AI systems to interact with humans to learn aligned and robust behaviors. The way robots and other AI systems behave is often the result of optimizing a reward function. However, manually designing good reward functions is highly challenging and error-prone, even for domain experts. Although reward functions are often difficult to manually specify, human feedback in the form of demonstrations or preferences is often much easier to obtain but can be difficult to interpret due to ambiguity and noise. Thus, it is critical that AI systems take into account epistemic uncertainty over the human's true intent. As part of the AAAI New Faculty Highlight Program, I will give an overview of my research progress along the following fundamental research areas: (1) efficiently quantifying uncertainty over human intent, (2) directly optimizing behavior to be robust to uncertainty over human intent, and (3) actively querying for additional human input to reduce uncertainty over human intent.

The future of artificial intelligence demands a paradigm shift toward multisensory perception—to systems that can digest ongoing multisensory observations, that can discover structure in unlabeled raw sensory data, and that can intelligently fuse useful information from different sensory modalities for decision-making. While we humans naturally perceive the world by looking, listening, touching, smelling, and tasting, traditional forms of machine intelligence mostly focus on a single sensory modality, particularly vision. Therefore, my research, which I refer to as multisensory machine intelligence, seeks to bridge this gap by empowering machines to emulate and enhance human capabilities in seeing, hearing, and feeling, ultimately enabling them to comprehensively perceive, understand, and interact with multisensory world.

The transfer of patients between two aircraft using an underway watercraft increases medical evacuation reach and flexibility in maritime environments. The selection of any one of multiple underway watercraft for patient exchange is complicated by participating aircraft utilization histories and participating watercraft positions and velocities. The selection problem is modeled as a semi-Markov decision process with an action space, including both fixed land and moving watercraft exchange points. Monte Carlo tree search with root parallelization is used to select optimal exchange points and determine aircraft dispatch times. Model parameters are varied in simulation to identify representative scenarios where watercraft exchange points reduce incident response times. We find that an optimal policy with watercraft exchange points outperforms an optimal policy without watercraft exchange points and a greedy policy by 35% and 40%, respectively. In partnership with the United States Army, we deploy for the first time the watercraft exchange point by executing a mock patient transfer with a manikin between two HH-60M medical evacuation helicopters and an underway Army Logistic Support Vessel south of the Hawaiian island of Oahu. Both helicopters were dispatched in accordance with our optimized decision strategy.

The advent of generative AI, particularly large language models like ChatGPT, has precipitated a seismic shift in academia. Far from a gradual evolution, its sudden emergence has jolted educational institutions, leaving many academics grappling with a perceived encroachment upon their intellectual domain. This upheaval has sparked intense debates, with concerns ranging from the erosion of academic integrity to the devaluation of scholarly labor. This essay contends that such apprehensions, while understandable, may overlook the transformative potential of AI as a collaborative tool. Drawing parallels to historical disruptions—such as the advent of photography challenging traditional art forms—we explore how AI can augment human creativity rather than supplant it. By examining the dynamics of authorship, originality, and accountability, we argue for a redefinition of these concepts in the context of AI-assisted work. Emphasizing the importance of human oversight in guiding AI outputs, we advocate for a framework that recognizes the symbiotic relationship between human intellect and machine efficiency. Such a perspective not only preserves the essence of academic rigor but also embraces the democratization of knowledge production. Ultimately, this essay calls for a balanced approach that mitigates risks while harnessing the innovative capacities of generative AI in academia.

The rapid advancement of generative artificial intelligence (AI) poses significant challenges to traditional copyright frameworks, intensifying debates over the copyrightability of AI-generated outputs. By comparing judicial practices in China and the United States, it has been observed that the United States maintains a conservative stance of adhering to substantive control, while China demonstrates an inclusive approach through the criterion of creative contribution. Building upon this, this article transcends the traditional binary judgment model and constructs a tiered copyright determination model. Based on the level of human control and contribution in the AI generation process, it introduces dimensions such as technological controllability and density of human intent, classifying generative AI into three tiers: strong protection, weak protection, and non-protection. Regarding the copyrightability of content generated by generative AI, this article argues that the issue should be addressed within the framework of copyright law itself. When human participation is involved and the substantial contribution of the direct user is reflected in the AI-generated content, meeting the requirements for copyrightable works under copyright law, corresponding protective measures should be granted.