相关参考文献
注意：仅列出部分参考文献，下载原文获取全部文献信息。
Article
Business, Finance
Michael Dowling et al.
Summary: Based on ratings from finance journal reviewers, the AI chatbot ChatGPT demonstrates significant usefulness in finance research, with potential for application in other domains. It excels in idea generation and data identification, while facing challenges in literature synthesis and developing appropriate testing frameworks. Moreover, the quality of output is influenced by the extent of private data and researcher domain expertise input. Ethical implications and other potential impacts are also considered.
FINANCE RESEARCH LETTERS
(2023)
Article
Computer Science, Theory & Methods
Pengfei Liu et al.
Summary: This article surveys and organizes research in prompt-based learning, a new paradigm in natural language processing. Prompt-based learning uses language models to directly model the probability of text, allowing for few-shot or zero-shot learning. The article introduces the basics of this paradigm, provides mathematical notations, and organizes existing work based on various dimensions. It also releases additional resources to make the field more accessible.
ACM COMPUTING SURVEYS
(2023)
Article
Computer Science, Theory & Methods
Bo Li et al.
Summary: The rapid development of AI technology has led to the deployment of various systems. However, many current AI systems have vulnerabilities, are biased, and lack privacy protection. This review provides a comprehensive guide for building trustworthy AI systems, introducing the framework of AI trustworthiness and organizing various approaches. Concrete action items are offered for practitioners and stakeholders to improve AI trustworthiness. Key opportunities and challenges for future development are identified, emphasizing the need for a paradigm shift towards comprehensively trustworthy AI systems.
ACM COMPUTING SURVEYS
(2023)
Article
Ethics
Yves Saint James Aquino et al.
Summary: This study examines strategies to mitigate algorithmic bias in healthcare AI and investigates the question of responsibility for bias. The findings reveal divergent views on bias as a problem, strategies to mitigate bias, and whether to include sociocultural identifiers in AI development. The study suggests interdisciplinary collaboration, tailored engagement activities, empirical studies, participatory methods, and increased diversity and inclusion as potential responses.
JOURNAL OF MEDICAL ETHICS
(2023)
Letter
Biochemistry & Molecular Biology
Yifan Peng et al.
Review
Medicine, General & Internal
Bishnu Bajgain et al.
Summary: This article aims to study the determinants of implementing AI-based clinical decision support tools through a scoping review, in order to help improve the quality of care, appropriate use of healthcare resources, and decrease healthcare provider burnout.
Article
Medicine, General & Internal
John W. Ayers et al.
Summary: The rapid expansion of virtual health care has led to an increase in patient messages and burnout among health care professionals. This study evaluated the ability of an AI chatbot assistant to provide quality and empathetic responses to patient questions, and found that the chatbot performed well in both aspects.
JAMA INTERNAL MEDICINE
(2023)
Article
Computer Science, Theory & Methods
Ziwei Ji et al.
Summary: Natural Language Generation (NLG) has made significant progress in recent years, thanks to deep learning technologies like Transformer-based language models. This advancement has resulted in more fluent and coherent NLG, leading to improvements in tasks such as summarization, dialogue generation, and data-to-text conversion. However, deep learning-based generation is prone to producing unintended text, which affects system performance and fails to meet user expectations in real-world scenarios. To address this issue, researchers have conducted studies on measuring and mitigating hallucinated texts. This survey provides an overview of the research progress and challenges in the hallucination problem in NLG, covering metrics, mitigation methods, and task-specific research progress.
ACM COMPUTING SURVEYS
(2023)
Article
Computer Science, Artificial Intelligence
Anton Sigfrids et al.
Summary: Human-centricity is considered a central aspect in AI development and governance, but current uses of Human-Centered AI (HCAI) risk downplaying promises of desirable, emancipatory technology. This article explores using the HCAI approach for technological emancipation in public AI governance and emphasizes the importance of expanding the traditional user-centered view to involve community- and society-centered perspectives.
FRONTIERS IN ARTIFICIAL INTELLIGENCE
(2023)
Article
Multidisciplinary Sciences
Karan Singhal et al.
Summary: This paper introduces a multi-domain benchmark for medical question answering, which evaluates the performance of models in terms of factuality, comprehension, reasoning, possible harm, and bias through human evaluation. In addition, it proposes instruction prompt tuning to align language models to new domains. The experimental results suggest the potential value of model scale and instruction prompt tuning in improving comprehension, knowledge recall, and reasoning abilities. The human evaluations reveal the limitations of current models and emphasize the importance of evaluation frameworks and method development in creating safe and helpful large language models for clinical applications.
Article
Radiology, Nuclear Medicine & Medical Imaging
Zhaoyi Sun et al.
Article
Urology & Nephrology
Giovanni E. Cacciamani et al.
Summary: Artificial intelligence (AI) is expected to revolutionize healthcare by improving the quality of care for patients and overcoming human fatigue barriers. In the field of oncology, AI has the potential to standardize the interpretation of radiological imaging, especially for prostate imaging. Combining AI with radiologist assessment shows superior performance in detecting prostate cancer, offering a hybrid system that maximizes patient care quality while reducing physician workload and burnout.
EUROPEAN UROLOGY OPEN SCIENCE
(2023)
Review
Surgery
Nithesh Naik et al.
Summary: The legal and ethical issues of Artificial Intelligence (AI) in healthcare include privacy, surveillance, bias, and the role of human judgment. Currently, there are no well-defined regulations in place to address these concerns, highlighting the need for algorithmic transparency, privacy protection, and cybersecurity measures.
FRONTIERS IN SURGERY
(2022)
Article
Computer Science, Information Systems
Julien Meyer et al.
Summary: This study aims to determine the reliance of pathologists on artificial intelligence (AI) and investigate whether providing information on AI influences this reliance. The results show that pathologists' accuracy is significantly higher with AI decision aids, and providing information on the algorithm does not significantly impact reliance. The study also finds that decisions are made faster when AI is provided.
JOURNAL OF THE AMERICAN MEDICAL INFORMATICS ASSOCIATION
(2022)
Article
Health Care Sciences & Services
Xi Yang et al.
Summary: This study develops a large clinical language model and evaluates it on five clinical NLP tasks. By scaling up the number of parameters and increasing the size of the training data, the model improves accuracy and shows potential for enhancing medical AI systems.
NPJ DIGITAL MEDICINE
(2022)
Editorial Material
Health Care Sciences & Services
Jakob Nikolas Kather et al.
Summary: The GLIDE model, a state-of-the-art text-to-image generative AI model, shows promising representations in cancer research and pathology, but lacks useful representations in radiology data.
NPJ DIGITAL MEDICINE
(2022)
Proceedings Paper
Computer Science, Information Systems
Mina Lee et al.
Summary: This paper discusses the potential and challenges of large language models (LMs) in language generation and proposes the use of curated interaction datasets for more insightful research on their generative capabilities. The COAUTHOR dataset is presented as an example to reveal GPT-3's abilities in creative and argumentative writing. This work contributes to a more principled discussion on the promises and pitfalls of language models in interaction design.
PROCEEDINGS OF THE 2022 CHI CONFERENCE ON HUMAN FACTORS IN COMPUTING SYSTEMS (CHI' 22)
(2022)
Proceedings Paper
Computer Science, Artificial Intelligence
Robin Rombach et al.
Summary: By applying diffusion models (DMs) in the latent space of powerful pretrained autoencoders, this paper achieves high-quality image generation on limited computational resources. The introduction of cross-attention layers further enhances the flexibility and performance of the generator.
2022 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR)
(2022)
Article
Health Care Sciences & Services
Jordan P. Richardson et al.
Summary: The study found that patients have various concerns regarding the applications of artificial intelligence in healthcare, such as safety, data privacy, and potential increases in costs. Patient acceptance of AI is contingent on mitigating these possible harms.
NPJ DIGITAL MEDICINE
(2021)
Article
Computer Science, Theory & Methods
Ishai Rosenberg et al.
Summary: This article presents a comprehensive summary of recent research on adversarial attacks against security solutions based on machine learning techniques, highlighting the associated risks. The methods of adversarial attacks are characterized based on occurrence stage, attacker goals, and capabilities, while categorizing the applications of attack and defense methods in the cyber security domain. It also discusses the impact of recent progress in adversarial learning fields on future research directions in cyber security.
ACM COMPUTING SURVEYS
(2021)
Proceedings Paper
Computer Science, Artificial Intelligence
Priyanka Ranade et al.
Summary: This paper demonstrates the automatic generation of fake CTI text descriptions using transformers for data poisoning attacks. The attacks result in negative impacts such as incorrect reasoning outputs and disruption of AI-based cyber defense systems.
2021 INTERNATIONAL JOINT CONFERENCE ON NEURAL NETWORKS (IJCNN)
(2021)
Review
Computer Science, Artificial Intelligence
Marcal Mora-Cantallops et al.
Summary: Traceability is considered a key requirement for trustworthy artificial intelligence, involving the need to maintain a complete account of the provenance of data, processes, and artifacts. However, a common approach and shared semantics are currently lacking in AI traceability tools, with some tools either not fully mature or falling into obsolescence, compromising research reproducibility.
BIG DATA AND COGNITIVE COMPUTING
(2021)
Article
Public, Environmental & Occupational Health
Ibrahim Habli et al.
BULLETIN OF THE WORLD HEALTH ORGANIZATION
(2020)
Article
Ophthalmology
Wei-Chun Lin et al.
TRANSLATIONAL VISION SCIENCE & TECHNOLOGY
(2020)
Review
Medical Informatics
Avishek Choudhury et al.
JMIR MEDICAL INFORMATICS
(2020)
Review
Health Care Sciences & Services
Maria Moudatsou et al.
Article
Computer Science, Artificial Intelligence
Adam Baker et al.
FRONTIERS IN ARTIFICIAL INTELLIGENCE
(2020)
Editorial Material
Multidisciplinary Sciences
Ravi B. Parikh et al.
Article
Multidisciplinary Sciences
Ziad Obermeyer et al.
Review
Radiology, Nuclear Medicine & Medical Imaging
Stephen Chan et al.
BRITISH JOURNAL OF RADIOLOGY
(2019)
Article
Computer Science, Information Systems
Anuj K. Dalal et al.
JOURNAL OF THE AMERICAN MEDICAL INFORMATICS ASSOCIATION
(2019)
Editorial Material
Medicine, General & Internal
Danton S. Char et al.
NEW ENGLAND JOURNAL OF MEDICINE
(2018)
Article
Multidisciplinary Sciences
Kelly M. Hoffman et al.
PROCEEDINGS OF THE NATIONAL ACADEMY OF SCIENCES OF THE UNITED STATES OF AMERICA
(2016)
Article
Behavioral Sciences
Joseph B. Lyons et al.
