Anthropic研究员Evan Hubinger在社交媒体上表示，人工智能在未来十年内有超过10%的概率"可能杀死所有人类"1。Hubinger指出，虽然当前AI模型的风险相对较低，但他担忧该技术可能快速自我改进至对人类构成威胁1。他坦言："我们真诚地相信AI对人类构成物种灭绝风险。我认为Anthropic在尽力，但我们还没有一个解决超级智能对齐的计划，也不清楚我们是否在正轨上。"1
Hubinger的警告发表之际，Anthropic因被指向英国AI安全研究所隐瞒其最新模型而受到质疑1。曾在Anthropic和OpenAI工作的Jacob Coxon也发文表达类似担忧，称这些系统将很快成为"超人类系统，可以黑进任何东西，一夜间革命化任何领域，并获得真实权力和资源"1。Anthropic 8月发布的安全报告称，对AI自动研究开发导致"灾难性伤害"的风险"信心较低"1。此外，OpenAI首席科学家Jakub Pachocki要求对AI进步保持"极度谨慎"1，而超过1300名AI公司员工签署公开信，呼吁美国政府支持国际努力"故意放缓前沿AI自动开发的步伐"1。
Evan Hubinger, a researcher at Anthropic, has expressed grave concerns about artificial intelligence, stating there is more than a 10% probability that AI "could kill all humans" within the next decade 1. Hubinger elaborated on his position, saying: "We genuinely believe AI poses an existential risk to humanity. I think Anthropic is trying, but we don't have a plan for solving superintelligent alignment, and it's unclear to me whether we're on track" 1. He cautioned that while current AI models present lower risks, the technology could rapidly self-improve to become catastrophically dangerous 1.
Hubinger's comments reflect broader concerns within the AI safety community. Jacob Coxon, who previously worked at both Anthropic and OpenAI, warned that advanced systems "will soon be superhuman, capable of hacking into anything, revolutionizing any field overnight, and acquiring real power and resources" 1. OpenAI's chief scientist Jakub Pachocki called for "extreme caution" regarding AI advancement 1. Additionally, 1,300 employees from AI companies signed an open letter urging the U.S. government to support international efforts to "deliberately slow the pace of frontier AI autonomous development" 1.
These warnings occur against scrutiny over AI company safety practices. Anthropic faced allegations of concealing its latest model from the United Kingdom's AI safety research institute 1, raising questions about corporate transparency in AI development. The company's August safety report stated it had "low confidence" in assessing the risk that AI automated research and development could cause "catastrophic harm" 1.
评论
还没有评论，欢迎留下第一条。