特雷弗·诺亚谈人工智能

qimuai 发布于 阅读:34 一手编译

特雷弗·诺亚谈人工智能

内容来源:https://www.geekwire.com/2026/etzioni-on-ai-trevor-noah-on-ai/

内容总结:

10月19日,特雷弗·诺亚在波士顿科学博物馆推出科学对话系列节目“Luminaries”,我与诺亚同台,由拉娜·埃尔·卡利乌比主持。拉娜是情感AI(读取人类情感的软件)领域的先驱,也是“AI先驱者”播客的主持人。她事后写道,她预料到诺亚会机智风趣,但“没想到他的人文情怀和思想深度如此令人震撼”。

以下是他的一些见解,关键表述保留原话。

AI看似深刻,实则只在表面运作
当拉娜在ChatGPT中搭建的AI媒人把诺亚选为她的匹配对象时,诺亚指出,这个AI对他们两人都知之甚少,只是从互联网上的信息拼凑出一个看似合理的答案。诺亚还提到,当你告诉AI它犯了一个错误时,它会感谢你,但往往并不改正错误:“这就像一个看了很多心理医生的朋友,却并没有真正改变。”

线下社群是抵御AI反乌托邦未来的防线
当被问及如何保护自己免受AI负面影响时,诺亚说,答案“要回到某种非常非技术、非常非AI的东西”。他引用罗伯特·帕特南的《独自打保龄》一书,指出社群形成的 places 会带来许多意想不到的好处,包括更负责任的政府。我反驳说,我父亲参与发起的1990年代社群主义运动已经失败了。诺亚回应说,我们把太多理念交给了知识分子,他们随后“套用了过多的学术框架”。社群主义之所以失败,他面无表情地说,“是因为它叫社群主义”——一个过于学术化的名字。他指出,第一代手机卖不动时没人说手机失败了,公司会继续推出新产品。他誓言永远不会停止对社群的投入。

诺亚对技术持乐观态度
尽管有担忧,诺亚“还没有见过哪一种技术爆发之后把所有人都甩在了后面”。汽车、电脑和互联网都普及到了普通人。AI已经达到了他所说的“工作智能”:它能做我们发明出来的工作,但我们一直在发明新的工作。盐和冰曾经很珍贵,如今餐馆免费提供。我们所看重的东西在变化,在诺亚看来,只有当我们用今天所看重的任务来定义“人”时,我们才是可被替代的。

真正的就业问题在于“跟车距离”
作为喜剧演员巡演时,诺亚看到柯达和汽车制造商斯图贝克撤离后,一些城镇变得空心化。工厂工人被告知自动化是进步。“白领工人现在正在经历许多蓝领工人几十年来经历的事情,”他说。他认为区别在于,白领离权力更近,所以报纸在报道。诺亚相信新工作会到来,就像电话交换员消失后那样。他担心的是中间的缺口。他用高速公路作类比,“跟车距离”是让你在前车之前停下来的空间。诺亚告诉科技领袖们,“身处特权位置时更容易有耐心”,并补充说,“但讽刺的是,他们在季度财报上可没有耐心。”

为什么AI抢不走诺亚的饭碗
ChatGPT走红后,节目主持人不断给他讲ChatGPT模仿诺亚风格写出的蹩脚笑话。他列举了AI在幽默方面举步维艰的四个原因:喜剧数据不够;模型是一个冻结的快照,而社会认为可接受的东西在不断变化;喜剧依赖于了解你的受众(有些笑话你知道好笑,但你不会讲给妈妈听);企业让模型变得过于规避风险。

职业保障
除了是《每日秀》前主持人、获奖作家和炙手可热的喜剧演员之外,诺亚还是微软的“首席提问官”。我提醒他毕加索关于电脑的名言——“它们没用,只会给你答案”——并开玩笑说,作为首席提问官,他拥有如今罕见的东西:职业保障。他笑了。当然,他真正的职业保障来自他的深度和人文情怀,这滋养了他非凡的幽默和机智。拉娜指出,每个人都在被AI取代,但诺亚本人就扮演过一个AI——在《黑豹:瓦坎达万岁》中为格里奥特配音。他打趣道:

中文翻译:

10月19日,特雷弗·诺亚在波士顿科学博物馆推出了一档名为“Luminaries”的科学对话系列节目。我与他同台,拉娜·埃尔·卡利乌比担任主持。拉娜是情感AI的先驱——这种软件能读取人类情感——同时也是“AI先驱”播客的主持人。她事后写道,她预料到特雷弗会机智风趣,但“没想到他的人文情怀和思想深度会如此令人震撼”。
以下是他的几点见解,原话加粗标出。
AI看似深刻,实则只在表面运作
拉娜在ChatGPT里搭建了一个AI相亲匹配器,当它把特雷弗选为她的匹配对象时,特雷弗指出,它对他们俩都知之甚少,不过是从互联网上的信息拼凑出了一个看似合理的答案。
特雷弗补充说,当你告诉AI它犯了错,它会感谢你,但往往并不会纠正错误:“就像一个朋友,做了很多次心理治疗,但实际上并没有什么改变。”
线下社群是抵御AI反乌托邦未来的防线
当被问及如何保护自己免受AI的负面影响时,特雷弗说答案“要回到某种非常非技术、非常非AI的东西”。他引用了罗伯特·帕特南的《独自打保龄》,指出在社群得以形成的地方,会积累许多意想不到的好处,包括更负责任的政府。
我反驳说,社群主义——我父亲在1990年代参与发起的培育社群的运动——已经失败了。特雷弗回应说,我们把太多理念交给了知识分子,他们于是“套用了太多学术理论”。他面无表情地打趣道,社群主义之所以失败,“是因为它叫社群主义”——一个过于学术化的名字。
他指出,第一代手机卖不动的时候,没人会说手机失败了;公司会不断重新发布。他誓言永不停止对社群的投入。
特雷弗对技术持乐观态度
尽管有所担忧,特雷弗“还没见过哪种技术爆发之后把所有人都甩在了后面”。汽车、电脑和互联网都普及到了普通人。
AI已经达到了他所说的“工作智能”:它能做我们发明出来的工作,但我们一直在发明新的工作。盐和冰曾经很珍贵,如今餐馆免费提供。我们所看重的东西在变化,而在特雷弗看来,只有当我们用今天所看重的任务来定义“人”时,我们才是可被替代的。
我们的对话转向了AI对就业的影响。
真正的就业问题是“跟车距离”
作为喜剧演员巡演时,特雷弗看到柯达和汽车制造商斯图贝克撤离后,一座座城镇被掏空。工厂工人被告知自动化是进步。“白领工人现在正在经历许多蓝领工人几十年来经历的事情,”他说。他认为区别在于,白领工人离权力更近,所以报纸在报道这件事。
特雷弗相信新工作会到来,就像电话交换员消失之后那样。让他担忧的是那段空档期。他用高速公路来类比——“跟车距离”就是让你在前方车辆前停下来的空间:
特雷弗告诉科技领袖们,“当你处于特权位置时,保持耐心会更容易,”他还说,“但讽刺的是,一到季度财报,他们就没有耐心了。”
为什么AI抢不走特雷弗的饭碗
ChatGPT走红之后,节目主持人不断给他讲一些蹩脚的笑话,那些笑话是ChatGPT拙劣模仿特雷弗风格写出来的。他一口气列出AI在幽默方面举步维艰的四个原因:喜剧数据不够;模型是一个冻结的快照,而社会认为可接受的东西在不断变化;喜剧依赖于了解你的受众(有些笑话你知道好笑,但你不会讲给你妈听);企业把模型做得过于保守。
职业保障
除了曾是《每日秀》主持人、获奖作家和炙手可热的喜剧演员之外,特雷弗·诺亚还是微软的“首席提问官”。
我提醒他毕加索关于计算机的那句名言——“它们毫无用处。它们只会给你答案”——并开玩笑说,作为首席提问官,他拥有如今稀罕的东西:职业保障。他笑了。当然,他真正的职业保障来自他的深度和人文情怀,正是这些滋养了他非凡的幽默与机智。
拉娜提到,每个人都在被AI取代,但特雷弗已经演过一个AI——在《黑豹:永远的瓦坎达》中为格里奥特配音。
他打趣道:

英文来源:

On Oct. 19, Trevor Noah launched Luminaries, a science conversation series at Boston’s Museum of Science. I joined him onstage, and Rana el Kaliouby moderated. Rana is a pioneer of emotion AI, software that reads human emotion, and the host of the Pioneers of AI podcast. She wrote afterward that she expected Trevor to be witty and funny, but “didn’t expect to be so blown away by his humanity and depth of ideas.”
Here are a few of his insights with his own words highlighted.
AI often appears deep but merely works on the surface
When an AI matchmaker Rana built in ChatGPT picked Trevor as her match, he pointed out that it knew very little about either of them. It assembled a plausible answer from what’s on the internet.
Trevor added that when you tell AI it made a mistake, it thanks you but often fails to correct the error: “It’s like a friend who’s been to a lot of therapy but doesn’t actually change.”
In-person community is a defense against a dystopian AI future
When asked how we protect ourselves from AI’s downsides, Trevor said the answer “goes back to something that is very un-technological and very un-AI.” He cited Robert Putnam’s book Bowling Alone, noting that places where communities form accrue many surprising benefits including more accountable governments.
I pushed back that communitarianism, the 1990s movement my father helped found to foster communities, had failed. Trevor responded that we hand too many ideas to intellectuals, who then “apply too much academia” to them. Communitarianism failed, he deadpanned, “because it’s called communitarianism.” An overly academic name.
He noted that nobody calls a phone a failure when the first version doesn’t sell; companies keep re-releasing it. He vowed never to stop investing in community.
Trevor is optimistic about technology
Despite his concerns, Trevor has “yet to see a technology that blew up and then left everyone behind.” Cars, computers and the internet all spread to ordinary people.
AI has reached what he calls “work intelligence”: it can do the jobs we invented, but we keep inventing new ones. Salt and ice used to be precious. Now restaurants hand them out for free. What we value changes, and in Trevor’s view, we’re expendable only if we define being human by the tasks we value today.
Our conversation turned to AI’s impact on jobs.
The real jobs question is the following distance
Touring as a comic, Trevor saw towns hollowed out after Kodak and the automaker Studebaker left. Factory workers were told automation was progress. “White-collar workers are now experiencing what many blue-collar workers experienced for decades,” he said. The difference, he argued, is that white-collar workers are closer to power, so newspapers are covering it.
Trevor is confident new jobs will come, as they did after telephone exchange operators disappeared. What worries him is the gap. His analogy is the highway, where the following distance is the space that lets you stop before you hit the car ahead:
Trevor tells tech leaders that “it’s easier to be patient when you are in a position of privilege,” and added, “But ironically, they don’t have patience when it comes to quarterly earnings.”
Why AI isn’t coming for Trevor’s job
After ChatGPT took off, show hosts kept telling him weak jokes that ChatGPT had written in a pale imitation of Trevor’s style. He rattled off four reasons AI struggles with humor: there isn’t enough comedy data, the model is a frozen snapshot while what society finds acceptable keeps shifting, comedy depends on knowing your audience (there are jokes you know are funny but you wouldn’t tell your mom), and companies have made models risk averse.
Job security
In addition to being the former host of The Daily Show, an award-winning author and a sought-after comedian, Trevor Noah is Microsoft’s Chief Questions Officer.
I reminded him of Picasso’s famous verdict on computers, “They are useless. They can only give you answers,” and joked that as Chief Questions Officer he has something rare these days: job security. He chuckled. Of course, his real job security comes from his depth and humanity, which feed his exceptional humor and wit.
Rana noted that everybody’s getting replaced by AI, but Trevor had played an AI, voicing Griot in Black Panther: Wakanda Forever.
He quipped:

Geekwire

文章目录


    扫描二维码,在手机上阅读