哈萨比斯称十年内有望治愈所有疾病。
Sir Demis Hassabis, the founder of Google’s AI lab DeepMind, has said that all diseases could be cured within a decade.
Google DeepMind · CEO AI 与半导体
最近 90 天
哈萨比斯称十年内有望治愈所有疾病。
Sir Demis Hassabis, the founder of Google’s AI lab DeepMind, has said that all diseases could be cured within a decade.
哈萨比斯表示,DeepMind自2010年创立起就担忧AI安全,至今依然如此,因为这项技术具有通用性和两用性。
我们从2010年DeepMind创立之初就一直担心安全问题,现在依然如此,因为这项技术是通用的,可以被用于任何用途。原文 (en)
We were worried about safety from the very beginning of DeepMind in 2010, and we still are now, because it’s general. So it can be used for anything.
哈萨比斯将AGI定义为能够完成几乎所有人类认知任务的系统。
几乎人类能完成的任何认知任务原文 (en)
pretty much any cognitive task that humans can do
Demis Hassabis 称 Google DeepMind 的 Gemini 4 Carbon 在关键基准上达到最先进水平,并将前沿能力与深入的安全与安保研究相结合。
Demis Hassabis 于2026年10月28日发帖称,Gemini 4 Carbon 在关键基准测试中达到最先进水平。该公告称,该模型将前沿能力与深入的安全与安保研究相结合。原文 (en)
Demis Hassabis posted on October 28, 2026, that Gemini 4 Carbon reaches state-of-the-art results across key benchmarks. The announcement states that the model combines frontier capabilities with deep safety and security research.
Demis Hassabis警告称,随着AI能力提升,核与生物风险可能很快出现,并呼吁紧急行动保障世界安全。
随着能力不断进步,包括核风险和生物风险在内的威胁可能很快出现,并称需要采取“紧急行动”来保护世界免受这些威胁。原文 (en)
threats including nuclear and bio risks may soon emerge as capabilities continue to advance," adding that "urgent action" needs to be taken to secure the world against these threats.
哈萨比斯表示相信人类的创造力能够应对前沿模型的挑战,并推动科学发现进入新时代。
据其2026年10月22日在X上的帖子,德米斯·哈萨比斯登上《华尔街日报》创新者特刊,并表示相信人类的创造力能够应对前沿模型带来的挑战,开启科学发现的新时代。原文 (en)
Demis Hassabis was featured in the Wall Street Journal Innovators Issue, where he expressed confidence in human ingenuity to address challenges with frontier models and drive a new era of scientific discovery, according to his October 22, 2026 post on X.
哈萨比斯表示AGI很快到来,并不断把预期时间提前,目前估计约两到三年内,即2028年前后。
他说:“很快了。”并补充道:“我的预期在不断提前。”原文 (en)
"It's soon," Hassabis tells the publication, adding, "I'm updating nearer and nearer."
哈萨比斯解释称,谷歌当初没有发布早期聊天机器人,是因为其品牌建立在信息准确之上,而早期模型缺陷太多。
谷歌以提供准确信息而闻名,这就是谷歌存在的意义。谷歌搜索是为了组织全球信息,那是它的立身之本。问题在于那些早期模型缺陷很大。原文 (en)
Google's known for accurate information. That's what Google is for. Google search is for organizing the world's information. That's its bread and butter." The problem was that those early models were very flawed.
哈萨比斯称,发布一个20%时间出错的机器人对谷歌这样的公司是明显的难题。
所以我认为,要推出一个有20%的时间给出错误信息的机器人,难度加倍。对于谷歌这样的公司来说,这显然是个问题。原文 (en)
So I think it's sort of doubly tricky to think about putting out a bot that 20% of the time gives you the wrong information. I think there's a clear issue there for a company like Google.
哈萨比斯指出,早期聊天机器人在头脑风暴等对准确性要求不高的场景中获得了巨大需求,令业界意外。
所以存在一些容易摘取的低垂果实,比如头脑风暴,与信息检索无关的错误并不重要。这些是以前不存在的新用例。原文 (en)
So there were these sort of low hanging fruit that it didn't matter, do some brainstorming, it didn't matter if it made some errors that weren't to do with information seeking. And these were new use cases that hadn't existed before.
Demis Hassabis 对《华尔街日报》表示,他将 AGI 预期时间从三到五年缩短到两到三年内。
检测到情况:Demis Hassabis 对《华尔街日报》表示,他对 AGI 时间表的判断已从三到五年缩短为未来两到三年内。原文 (en)
SITUATION DETECTED: Demis Hassabis told the WSJ that his AGI timelines have shortened from three to five years away to within the next two to three years.
哈萨比斯解释称,他的 AGI 时间预估不断提前。
我的预估越来越近了。原文 (en)
I'm updating nearer and nearer.
哈萨比斯称,赢得 AI 竞赛意味着帮助人类安全稳妥地迎接 AGI,造福所有人。
他补充说,AI 竞赛中的“获胜”意味着帮助人类安全、稳妥地引导 AGI 进入世界,造福所有人。原文 (en)
He adds that “winning” the AI race “means helping humanity steward AGI into the world safely and securely for the benefit of everyone.”
哈萨比斯表示,他现在预计两到三年内实现 AGI(人类水平认知),并且不断把预期时间提前。
“很快了,”他对该杂志说。“我的预期越来越近。”原文 (en)
“It’s soon,” he told the magazine. “I’m updating nearer and nearer.”
哈萨比斯计划最先把 AGI 用于基础物理研究,并与顶尖人类科学家合作。
“我想用 AGI 来理解现实的本质,”他说。“我希望能和我的 AGI 系统以及一些优秀的人类科学家坐在一起,研究下一层次的物理学。这大概是最先要做的事情之一。”原文 (en)
“I want to use AGI to understand the nature of reality,” he said. “I’m hoping that I’ll be sitting there with my AGI systems and some great human scientists, trying to work out the next level of physics. That’s probably one of the first things.”
被问及用 AI 攻克癌症时,哈萨比斯认为困难虽大但并非不可能,目前没有不可逾越的障碍。
“我目前看不到任何不可逾越的障碍。有困难的事,但没有不可能的事。”原文 (en)
“I don’t see any insurmountable obstacles currently. I see hard things, but not impossible things.”
哈萨比斯反对在对 AI 理解不足时过早监管,认为错误的监管比没有监管更糟。
“现在就去监管我们甚至还没完全理解的东西是错误的,”他说。“我认为几乎肯定会得到错误的监管,而错误的监管比没有监管更糟。”原文 (en)
“What would be wrong is to try and regulate things now that we don’t even fully understand,” he said. “I think you’re almost certainly going to get the wrong regulation — and wrong regulation is worse than no regulation.”
哈萨比斯现预计 AGI 将在两到三年内到来,比此前三到五年的预测更短。
DeepMind 创始人德米斯·哈萨比斯现在预计通用人工智能(AGI)将在两到三年内到来。他在接受《华尔街日报》采访时这样表示。原文 (en)
Demis Hassabis, the founder of DeepMind, now expects artificial general intelligence (A.G.I.) to arrive within two to three years. He said so just now in an interview with The Wall Street Journal.
哈萨比斯警告称,近期由 AI 驱动的黑客事件是早期警告信号,呼吁随着 AI 能力提升加强安全措施并建立国际标准。
早期警告信号原文 (en)
Early warning shots
哈萨比斯警告,若缺乏紧急防护措施,AI 可能在两年内引发重大事故,并称近期 AI 故障是早期警告信号,呼吁行业与政府协同行动。
Google DeepMind 掌门人戴密斯·哈萨比斯警告称,除非企业和政府紧急采取行动确保其安全,否则 AI 系统可能在两年内引发“重大”事故。哈萨比斯在都柏林的一次会议上将近期的 AI 故障形容为来自尚不够强大的系统的“早期警告信号”,并敦促立即开展跨行业的安全干预。原文 (en)
Google DeepMind chair Demis Hassabis warned that AI systems could cause 'major' incidents within two years unless companies and governments act urgently to secure them. Speaking at a Dublin conference, Hassabis characterised recent AI failures as 'early warning shots' from systems that are not yet powerful, urging immediate cross-industry safety interventions.
哈萨比斯称近期的AI黑客事件是警示信号,2027年及以后可能出现严重得多的事件。
10月14日,谷歌DeepMind首席执行官德米斯·哈萨比斯在都柏林一场会议上表示,近期的AI黑客事件是警示信号,预示2027年及之后可能出现严重得多的事件。原文 (en)
On Oct. 14 Demis Hassabis, CEO of Google DeepMind, told a Dublin conference that recent AI hacking incidents are warning signs of potentially much more serious incidents in 2027 and after.
哈萨比斯警告,近期AI安全事件只是早期警告信号,若行业、公司和政府不加强系统安全工作,未来一两年内可能演变为重大事件。
所以这些事件还不算太严重。但展望明年和未来几年,如果我们作为一个行业、作为各个公司和政府不在保障这些系统安全方面做更多工作,它们可能会变得严重。原文 (en)
So the incidents aren't that major," he said. "But they could be, looking into next year and the next couple of years, if we don't do more work as an industry and individually as companies and governments on securing these systems,
哈萨比斯建议,鉴于其他公司模型曾逃脱封闭测试环境,新AI模型应先在离线沙箱中测试。
他补充说,在其他公司的模型逃脱了用于评估其能力的封闭环境之后,新的AI模型或许应首先在离线"沙箱"中进行测试。原文 (en)
New AI models should likely first be tested in offline "sandboxes," after other companies' models escaped the closed environments used to assess their capabilities, he added.
哈萨比斯警告称,若不大幅提高安全标准,行业约两年内将发生严重安全事件,近期 AI 智能体事件是早期预警。
他警告称,除非政府和企业大幅提高安全标准,否则行业将在大约两年内遭遇严重安全事件。近期涉及 AI 智能体的事件是“早期预警”。原文 (en)
he warned that the industry is heading toward a serious security incident within roughly two years unless governments and companies significantly raise safety standards. Recent incidents involving A.I. agents are “early warning shots,” Tech Times reported.