据美国科技媒体《WIRED》报道，当好莱坞各大制片公司忙着将生成式AI引入电影和电视剧制作时，一群专门拿AI开玩笑的内容创作者却意外走红。他们通过戏弄AI面试官、揭露ChatGPT的低级错误，吸引数百万粉丝，甚至获得了好莱坞经纪公司的签约。
这些创作者的成功，恰恰建立在AI的不完美之上。
一本正经胡说八道，AI面试官照单全收
在一段走红的短视频中，美国喜剧创作者格雷厄姆·拉贝（Graham Raabe）正在接受一位AI面试官的求职面试。
面对虚拟面试官，拉贝自称拥有10年销售经验，擅长管理一些完全不存在的业务，还骄傲地介绍自己掌握了名为“goosebuckle”的虚构技能。
令人啼笑皆非的是，AI面试官不仅没有发现异常，反而认真回应：“管理badger margin听起来是一项非常重要的技能，你在goosebuckle方面的经验也令人印象深刻。”
当被问及期望薪资时，拉贝随口回答“600”。AI却将其理解为60万美元年薪，并称赞他的自信。
整场面试中，无论拉贝说出多么荒谬的话，AI始终保持礼貌、积极和赞赏。
这正是拉贝的喜剧套路：利用AI过度迎合用户、缺乏常识判断的特点，制造一本正经的荒诞效果。
拉贝原本是一家金融科技公司的销售团队经理，业余时间从事脱口秀和剧本创作。今年5月，他开始在Instagram和TikTok发布职场喜剧视频，随后逐渐转向AI题材。
他推出的《如何搞砸你的AI面试》系列迅速走红。过去两个月，他的Instagram粉丝从约5万人增长至超过10万人，并获得经纪公司Dulcedo Management的签约。
拉贝将AI称为即兴喜剧的“终极捧哏”。
靠ChatGPT犯错，吸引200万粉丝
拉贝并非个例。
另一位喜剧创作者马特·哈斯克（Matt Husk）则专门拍摄自己与ChatGPT的荒诞对话。
他经常向ChatGPT提出看似简单的问题，例如让它从1数到100，或者帮助解决陷入流沙的困境。
结果，AI经常把事情搞砸。
更有意思的是，当哈斯克指出错误时，ChatGPT有时不仅不承认，还会试图解释自己为什么没有错。
今年4月，在科技播客《Mostly Human》中，OpenAI CEO萨姆·奥特曼（Sam Altman）观看了一段哈斯克的视频。
视频里，哈斯克要求ChatGPT充当秒表，结果AI把短短几秒钟误判成了“大约10分钟”。
奥特曼看到后笑了起来，承认这是一个已知问题，并预测语音模型可能还需要一年才能准确计时。
靠着不断展示AI的荒谬表现，哈斯克在Instagram、TikTok和YouTube累计吸引约200万粉丝，并与人才经纪公司Unicorn签约。
这些视频不仅为创作者带来流量，也在客观上成为AI产品的公开测试。
拉贝透露，一些AI视频虚拟人产品经理甚至会主动联系他，感谢他发现产品的能力缺陷，因为这让开发团队知道应该修复什么问题。
模仿AI说话，也能成为一门生意
还有一类创作者，干脆自己扮演AI。
Instagram账号@ahh.gpt的创作者萨姆·瓦塞夫（Sam Wassef），通过模仿聊天机器人的说话方式走红。
他会刻意使用AI常见的空洞表达、突兀转折和不合逻辑的回答，有时还会在表演过程中大量喝水，营造荒诞效果。
在其中一段视频中，他不断谈论AI将带来的美好未来，但镜头逐渐拉远，观众才发现他独自站在一座几乎空无一人的购物中心，周围全是关闭的商铺。
这种带有讽刺意味的内容，让他的Instagram粉丝超过86万。
如今，瓦塞夫也已经与Night Media旗下的经纪人签约。
好莱坞为何看上这些“AI黑粉”？
Dulcedo Management游戏与娱乐业务副总监亚历克斯·梅伦德斯（Alex Melendez）认为，这类内容走红并不令人意外。
“AI正在改变行业、取代工作岗位的说法随处可见，但你转头却发现，它连一些极其基础的事情都做不好。”
他认为，这种巨大反差天然具有喜剧效果。
更重要的是，这些创作者让公众能够用幽默的方式理解AI带来的变化，而不是仅仅感到恐惧或焦虑。
对好莱坞经纪公司来说，他们看中的也不仅是AI话题本身，而是创作者能否将复杂技术转化成大众愿意观看和分享的内容。
不过，梅伦德斯也提醒，依赖某一种热门题材走红并不意味着能够长期成功。
经纪公司最终希望观众喜欢的是创作者本人，而不只是他们使用的内容形式。
“病毒式传播可以打开大门，但真正的才华才能建立持久的职业生涯。”
从AI面试官到ChatGPT，从虚拟人到语音助手，生成式AI的快速普及正在为喜剧创作者提供源源不断的素材。
一个颇具讽刺意味的现象正在出现：当科技公司投入巨资让AI变得越来越像人类时，另一群人却靠着展示AI有多么不像人类，找到了新的职业机会。
本文由 AI图鉴自动抓取并翻译，版权归原作者所有；如需查看原文请点下方链接。
According to WIRED, a group of content creators who joked about AI unexpectedly lost when major Hollywood producers were busy introducing the generation of AI into film and television productions. They have attracted millions of fans, even contracted by Hollywood Brokers, by mocking the interview officers in AI and exposing ChatGPT ' s low-level mistakes.
The success of these creators is based on the imperfections of AI.
I'll take it all.
In a popular short video, the American comedy creator Graham Raabe is being interviewed for employment by an AI interviewer.
Facing a virtual interviewer, Rabe claims to have 10 years of sales experience, is good at managing a number of completely non-existent operations and is proud to present a fictional skill known as “goosebuckle”.
Paradoxically, not only did the AI interviewer not find out about the anomaly, but he responded seriously: “Managing badger margin sounds like a very important skill, and your experience with goosebuckle is impressive.”
When asked about expected pay, Labbé responded “600”. AI understood it as a $600,000 annual salary and praised his confidence.
Throughout the interview, no matter how ridiculous Rabe ' s words were, AI remained polite, active and appreciated.
This is the comedy routine of Rabe: to create a serious and absurd effect by using the features of AAI that are overly user-friendly and lack common sense judgement.
Labbé was originally the sales team manager of a financial technology company, and was engaged in talk shows and script creations in amateur hours. In May this year, he began a comedy video at Instagram and TikTokpublish in the workplace, and then gradually moved to the subject of AI.
He launched the "How to screw up your AI interview" series, which quickly went red. Over the past two months, his Instagram fan has grown from about 50,000 to over 100,000 and has been contracted by the broker, Dulcedo Management.
Rabe called AI the "end of all" comedy.
It's a mistake to make a mistake.
Rabe is not an example.
Another comedian, Matt Husk, specializes in his own ridiculous conversation with ChatGPT.
He often raises with ChatGPT seemingly simple issues, such as allowing it to count between 1 and 100, or helping to resolve the plight of falling into the sand.
As a result, AI often screws up.
More interestingly, when Haskell points to mistakes, ChatGPT sometimes not only denies them, but also tries to explain why he is right.
In April this year, Open AI CEO Sam Altman, a technology podcast, viewed a Haskell video.
In the video, Haskell asked ChatGPT to act as a watch, resulting in AI miscalculating "about 10 minutes" for a few short seconds.
Otman smiled when he saw it, admitted that it was a known problem and predicted that it might take another year for the voice model to count accurately.
Based on the ongoing display of AI's absurdity, Haskell has accumulated around 2 million fans in Instagram, TikTok and YouTube and signed with the talent broker Unicorn.
These videos not only bring traffic to creators, but also objectively become open tests of AI products.
Labbé revealed that some AI video virtuals managers would even contact him on their own initiative to thank him for discovering that the product was deficient, as it gave the development team an idea of what should be fixed.
Imitating AI can be a business.
There's another kind of creator, just play AI.
Instagram account @ahh.gpt's creator, Sam Wassef, went red by imitating chat robots.
He will deliberately use the usual empty expression of AI, sudden turns and illogical answers, and sometimes drink a lot of water during performances to create absurd effects.
In one of the videos, he kept talking about the future that AI would bring, but shot was moving away and the audience found him standing alone in an almost empty shopping mall surrounded by closed shops.
The irony is that he has more than 860,000 Instagram fans.
Today, Vazev has also contracted with the broker under the banner of Night Media.
Why does Hollywood like "AI black powder"?
Alex Melendez, Deputy Director of Play and Recreation, Dulcedo Management, considered that the red-hotness of such content was not surprising.
“Ai is changing the industry to replace the job, and you turn around and you find that it can't even do very basic things.”
In his view, such a huge contrast had natural comedy effects.
More importantly, these creators allow the public to understand the changes brought about by AI in a humorous manner, rather than simply feeling fear or anxiety.
For Hollywood Brokers, it is not just AI itself that they see, but rather the ability of creators to translate complex technologies into content that the public is willing to see and share.
Meléndez, however, also cautioned that relying on some sort of hot topic does not mean long-term success.
In the end, brokerage companies want the audience to like the creators themselves, not just the content they use.
“Victoric transmission can open doors, but true talent can create a lasting career.”
From AI interviewer to ChatGPT, from virtual to voice assistant, the rapid spread of generating AI is a constant source of asset for comedy creators.
An ironic phenomenon is emerging: when technology companies invest heavily in making AI more and more human, another group of people find new career opportunities by showing how unlike they are.
Auto-fetched and translated by AI Atlas. All rights belong to the original author.