MIT Technology Review发布的《The Download》简报汇总了近期科技热点,核心议题包括对六项曾承诺改变世界的技术的进展评估以及关于“认知永生”的技术探索 1。该系列报道具体审视了核动力飞船、药丸式运动、种间交流、固态电池、清理PFAS以及消除交通死亡这六大领域的实际发展状况 1。此外,一场围绕“认知永生”主题的圆桌会议定于11月17日(周二)举行 1。
在产业与政策层面,Nvidia首席执行官黄仁勋考虑重返CEO职位;埃隆·马斯克加入了呼吁召回他的签名行动;特朗普计划将Nvidia命名为AI“国家冠军” 1。SpaceX首次成功捕获从轨道返回的火箭 1。国际方面,十个国家同意合作保护AI供应链免受中国影响 1。医疗领域迎来突破,首个体内CRISPR疗法预计明年到达患者手中,最早3月获得FDA批准 1。同时,美国国会计划通过法案让数据中心为电力支付数十亿美元额外费用 1。Apple计划明年推出首款智能眼镜 1。Bill Gates通过投资垃圾填埋场运营商获利290亿美元 1。
针对人工智能治理与数据质量,前OpenAI董事会成员Helen Toner表示:“如果我们对未来的希望寄托于任何一个个人的可信度上,那我们就完蛋了。” 1 此外,格陵兰语维基百科大量内容由机器翻译生成且存在错误,这可能威胁依赖该数据训练的小语种AI模型 1。
MIT Technology Review has released an edition of its daily newsletter, "The Download," which examines six technologies previously promised to reshape the world and explores the emerging concept of "cognitive immortality." 1 A dedicated Roundtables event focusing on cognitive immortality is scheduled for Tuesday, November 17. 1
The newsletter highlights several major developments in the technology sector, including SpaceX’s first successful capture of a returning rocket from orbit and an agreement among ten nations to collaborate on protecting AI supply chains from Chinese influence. 1 In healthcare, the first in-body CRISPR therapy is expected to reach patients next year, with potential FDA approval as early as March. 1 Meanwhile, political dynamics surrounding Nvidia continue to evolve; CEO Jensen Huang is considering a return to his role, with Elon Musk joining a petition to recall him and former President Trump planning to designate Nvidia as an AI "national champion." 1
Legislative and corporate moves also feature prominently, as the U.S. Congress proposes a bill requiring data centers to pay billions of dollars in additional electricity costs, while Apple plans to launch its first smart glasses next year. 1 Bill Gates has reportedly gained $29 billion by investing in landfill operators. 1 Helen Toner, a former OpenAI board member, emphasized the risks of relying on individual credibility for future hopes, stating, "If we place our hopes for the future on the credibility of any one person, we are doomed." 1 Additionally, concerns have been raised about the quality of AI training data, specifically noting that much of the Greenlandic Wikipedia content is machine-translated and error-prone, potentially impacting small-language AI models. 1
评论
还没有评论,欢迎留下第一条。