具有计算生物学背景的研究人员撰文反驳了经济学家Noah Smith关于AI可能设计致命超级病毒的观点1。作者基于病毒学专业知识论证，这类风险在现实中极为渺茫1。
作者指出，病毒的传播能力与致命性之间存在根本性权衡1。以麻疹为例，虽然传播力极强，但致死率仅为千分之一；而H5N1禽流感致命性高，却难以在人群中有效传播1。相比之下，COVID-19作为自然病毒，兼具中等致命性、高传播力和无症状传播特征，被认为已经接近"理想"大流行病毒的特征1。即便利用最先进的AI蛋白质设计系统，博士生设计简单肽结合剂也需耗费数月至数年时间，且多数设计最终失败1。
作者认为，真正应当警惕的是现实已在发生的病毒威胁1。当前刚果正遭遇大规模埃博拉疫情，美国麻疹病例激增，撒哈拉以南非洲HIV广泛传播，但这些真实风险往往受关注不足1。作者呼吁公众将焦点转向实际的病毒风险和公共卫生实践，而非陷入对虚构AI病毒的恐慌1。
A researcher with expertise in computational virology and artificial protein design has challenged widespread concerns about AI-designed pathogens, arguing that creating a lethal, highly transmissible virus through machine learning remains extraordinarily difficult 1.
The expert, drawing on experience developing AI protein design systems, contends that even advanced computational tools require months to years for PhD students to design simple peptide binders, with most attempts failing 1. The fundamental constraint, according to the analysis, lies in an inherent trade-off between viral transmissibility and lethality: measles spreads extremely effectively but kills only one in a thousand infected individuals, while H5N1 avian flu proves highly fatal but spreads poorly among humans 1. COVID-19, a naturally occurring virus, is characterized as approaching an "ideal" pandemic pathogen by combining moderate mortality, easy transmission, and asymptomatic spread—yet even this virus emerged through natural evolution rather than design 1.
Rather than fearing hypothetical AI-engineered pathogens, the researcher emphasizes that genuine public health threats already demand attention 1. Active crises include massive Ebola outbreaks in the Democratic Republic of Congo, rising measles cases in the United States, and widespread HIV transmission in sub-Saharan Africa, despite receiving limited public focus 1. The author concludes that concern should center on real viral risks and evidence-based public health practices rather than what is characterized as "fictional AI-designed virus" panic 1.
评论
还没有评论，欢迎留下第一条。