试论生成式人工智能的医疗应用能力与风险边界
作者简介:
王硕（1999-），男，博士研究生，研究方向：科技伦理
王硕（1999-），男，博士研究生，研究方向：科技伦理
通讯作者: 刘瑶瑶（1993-），女，博士，助理研究员，研究方向：医学人类学、科技与社会。E-mail：sanfy611@mail.tsinghua.edu.cn
Exploring the Capabilities and Risk Delineation of Generative AI in Medical Applications
-
摘要: 生成式人工智能医疗应用的伦理探讨需要对两个基本问题进行谨慎、清晰的划界。一方面，“能力”的划界旨在澄清生成式人工智能医疗应用的核心优势与技术局限，基于属种差异，明确生成式人工智能医疗应用相较于传统医疗人工智能的关键差异，避免陷入“ChatGPT全能论”或“ChatGPT无能论”的窠臼。另一方面，“风险”的划界旨在区分生成式人工智能医疗应用带来的问题类别及特征，厘清不同风险的独特性质与根源，包括传统人工智能的共有风险、生成式人工智能的特有风险以及生成式人工智能同医疗实践相结合的特性风险，推动风险的分类管理并推进伦理治理的不断完善。Abstract: Ethical discussions on the application of generative artificial intelligence (AI) in healthcare require careful and clear delineation of two fundamental issues. On one hand, the delineation of "capabilities" aims to clarify the core advantages and technical limitations of generative AI in medical applications. Based on species differences, it is essential to delineate the key differences between generative AI medical applications and traditional medical AI, avoiding falling into the trap of "ChatGPT omnipotence" or "ChatGPT impotence". On the other hand, the delineation of "risks" aims to distinguish the categories and characteristics of problems brought about by generative AI medical applications, clarifying the unique nature and roots of different risks. This includes common risks of traditional AI, unique risks of generative AI, and characteristic risks associated with the integration of generative AI with medical practice. This promotes the classification and management of risks and advances the continuous improvement of ethical governance.
-
-
[1] 梅景曌, 高艳红. ChatGPT在检验医学中的应用挑战[J]. 标记免疫分析与临床, 2023, 30(4): 695-698.
[2] DAVE T, ATHALURI S A, SINGH S. ChatGPT in medicine: An overview of its applications, advantages, limitations, future prospects, and ethical considerations[J]. Front Artif Intell, 2023, 6: 1169595.
[3] KHAN R A, JAWAID M, KHAN A R, et al. ChatGPT: Reshaping medical education and clinical management[J]. Pak J Med Sci, 2023, 39(2): 605-607.
[4] PATEL S B, LAM K. ChatGPT: The future of discharge summaries[J]. Lancet Digit Health, 2023, 5(3): e107-e108. doi: 10.1016/S2589-7500(23)00021-3
[5] BETZLER B K, CHEN H, CHENG C Y, et al. Large language models and their impact in ophthalmology[J]. Lancet Digit Health, 2023, 5(12): e917-e924. doi: 10.1016/S2589-7500(23)00201-7
[6] 康砚澜, 郭倩宇, 张文强, 等. 基于知识增强的医学语言模型: 现状、技术与应用[J]. 医学信息学杂志, 2023, 44(9): 12-22. doi: 10.3969/j.issn.1673-6036.2023.09.002
[7] LI Y, LI Z, ZHANG K, et al. ChatDoctor: A medical chat model fine-tuned on a large language model meta-AI (LLaMA) using medical domain knowledge[J]. Cureus, 2023, 15(6): e40895.
[8] LIEBRENZ M, SCHLEIFER R, BUADZE A, et al. Generating scholarly content with ChatGPT: Ethical challenges for medical publishing[J]. Lancet Digit Health, 2023, 5(3): e105-e106. doi: 10.1016/S2589-7500(23)00019-5
[9] 肖仰华, 徐一丹. 大规模生成式语言模型在医疗领域的应用: 机遇与挑战[J]. 医学信息学杂志, 2023, 44(9): 1-11.
[10] 李正风, 刘瑶瑶. 科技伦理治理要准确把握新科技革命及其伦理问题的新特点[J]. 科学通报, 2024, 69(13): 1677-1680.
[11] DIGIORGIO A M, EHRENFELD J M. Artificial intelligence in medicine & ChatGPT: De-tether the physician[J]. J Med Syst, 2023, 47(1): 32. doi: 10.1007/s10916-023-01926-3
[12] CADAMURO J, CABITZA F, DEBELJAK Z, et al. Potentials and pitfalls of ChatGPT and natural-language artificial intelligence models for the understanding of laboratory medicine test results: An assessment by the European Federation of Clinical Chemistry and Laboratory Medicine (EFLM) Working Group on Artificial Intelligence(WG-AI)[J]. Clin Chem Lab Med, 2023, 61(7): 1158-1166. doi: 10.1515/cclm-2023-0355
[13] CHENG K, SUN Z, HE Y, et al. The potential impact of ChatGPT/GPT-4 on surgery: Will it topple the profession of surgeons?[J]. Int J Surg, 2023, 109(5): 1545-1547. doi: 10.1097/JS9.0000000000000388
[14] MOOR M, BANERJEE O, SHAKERI Z, et al. Foundation models for generalist medical artificial intelligence[J]. Nature, 2023, 616(7956): 259-265. doi: 10.1038/s41586-023-05881-4
[15] ALI S R, DOBBS T D, HUTCHINGS H A, et al. Using ChatGPT to write patient clinic letters[J]. Lancet Digit Health, 2023, 5: e179-e181. doi: 10.1016/S2589-7500(23)00048-1
[16] CASCELLA M, MONTOMOLI J, BELLINI V, et al. Evaluating the feasibility of ChatGPT in healthcare: An analysis of multiple clinical and research scenarios[J]. J Med Syst, 2023, 47(1): 33. doi: 10.1007/s10916-023-01925-4
[17] ŠLAPETA J. Are ChatGPT and other pretrained language models good parasitologists?[J]. Trends Parasitol, 2023, 39(5): 314-316. doi: 10.1016/j.pt.2023.02.006
[18] SANTO D S E, JOVIANO-SANTOS J V. Exploring the use of ChatGPT for guidance during unexpected labour[J]. Eur J Obstet Gynecol Reprod Biol, 2023, 285: 208-209.
[19] 陈小平. 大模型关联度预测的形式化和语义解释研究[J]. 智能系统学报, 2023, 18(4): 894-900. doi: 10.11992/tis.202306045
[20] SINGHAL K, AZIZI S, TU T. et al. Large language models encode clinical knowledge[J]. Nature, 2023, 620(7972): 172-180. doi: 10.1038/s41586-023-06291-2
[21] World Health Organization. Ethics and governance of artificial intelligence for health: Guidance on large multi-modal models[EB/OL]. (2024-01-18)[2024-03-06]. https://www.who.int/publications/i/item/9789240084759.
World Health Organization. Ethics and governance of artificial intelligence for health: Guidance on large multi-modal models[EB/OL]. (2024-01-18)[2024-03-06]. https://www.who.int/publications/i/item/9789240084759.
[22] 颜见智, 何雨鑫, 骆子烨, 等. 生成式大语言模型在医疗领域的潜在典型应用与面临的挑战[J]. 医学信息学杂志, 2023, 44(9): 23-31. doi: 10.3969/j.issn.1673-6036.2023.09.003
[23] 王 硕, 文侃骁, 刘天语, 等. 医疗机器人的伦理风险与治理探析[J]. 医学与哲学, 2023, 44(23): 16-21.
[24] 汪 琛. 医疗人工智能伦理治理的问题、困境与求解[J/OL]. 科学学研究, 1-18(2024-03-13)[2024-03-31]. https://doi.org/10.16192/j.cnki.1003-2053.20240313.001.
[25] 刘天语, 王 硕, 刘鸿宇. “算法制”伦理: 数字社会的组织伦理新向度[J]. 自然辩证法研究, 2023, 39(6): 78-84.
[26] HARRIS E. Large language models answer medical questions accurately, but can't match clinicians' knowledge[J]. JAMA, 2023, 330(9): 792-794. doi: 10.1001/jama.2023.14311
[27] RAO A, PANG M, KIM J, et al. Assessing the utility of ChatGPT throughout the entire clinical workflow: Development and usabrlity study[J]. J Med Internet Res, 2023, 25: e45659.
[28] BISWAS S. ChatGPT and the future of medical writing[J]. Radiology, 2023, 307(2): e223312. doi: 10.1148/radiol.223312
[29] THIRUNAVUKARASU A J, TING D S J, ELANGOVAN K, et al. Large language models in medicine[J]. Nat Med, 2023, 29(8): 1930-1940. doi: 10.1038/s41591-023-02448-8
[30] 陈小平. 大模型: 从基础研究到治理挑战[J]. 中国人工智能学会通讯, 2024, 14(1): 2-9.
[31] FERRES J M L, WEEKS W B, CHU L C, et al. Beyond chatting: The opportunities and challenges of ChatGPT in medicine and radiology[J]. DiagnInterv Imaging, 2023, 104(6): 263-264.
[32] KUNG T H, CHEATHAM M, MEDENILLA A, et al. Performance of ChatGPT on USMLE: Potential for AI-assisted medical education using large language models[J]. PLoS Digit Health, 2023, 2(2): e0000198.
[33] BRIN D, SORIN V, VAID A, et al. Comparing ChatGPT and GPT-4 performance in USMLE soft skill assessments[J]. Sci Rep, 2023, 13(1): 16492. doi: 10.1038/s41598-023-43436-9
下载:
图(2)
计量
- 文章访问数: 934
- PDF下载数: 410
- 施引文献: 0