AI-Generated Jokes Resemble 'Cruise Ship Comedy Material From The 1950s' — Google DeepMind Experiment Turns Out To Be A Mixed Bag
AI-Generated Jokes Resemble 'Cruise Ship Comedy Material From The 1950s' — Google DeepMind Experiment Turns Out To Be A Mixed Bag
Alphabet Inc.'s (NASDAQ:GOOG) (NASDAQ:GOOGL) Google DeepMind researchers have conducted a study on the comedic capabilities of artificial intelligence. The results, presented at the ACM FAccT conference, were a mixed bag, with AI struggling to generate original and engaging comedy content.
Alphabet Inc.的（納斯達克：GOOG）（納斯達克：GOOGL）Google DeepMind研究人員對人工智能的喜劇能力進行了研究。結果在ACM FAccT會議上發佈，是個半途而廢的結果，因爲人工智能難以生成原創和有吸引力的喜劇內容。
What Happened: The study, led by Piotr Mirowski, a researcher at Google DeepMind and part-time improv comedian, involved 20 professional comedians who were already using AI in their work.
發起研究的是Google DeepMind的研究員Piotr Mirowski，同時也是兼職即興喜劇演員。這項研究涉及到20名職業喜劇演員，他們已經在工作中使用人工智能。
The comedians were asked to use large language models (LLMs) like OpenAI's ChatGPT or Google Gemini to generate comedic material, according to an MIT technology review.
這些喜劇演員被要求使用OpenAI的ChatGPT或Google Gemini等大型語言模型(LLMs)生成喜劇材料，根據麻省理工技術評論。
While the AI models were found useful for structuring monologues or producing rough drafts, they fell short in creating original, engaging, or funny content. The AI-generated jokes were described as bland and generic, with one comedian comparing them to "cruise ship comedy material from the 1950s."
雖然人工智能模型被發現在構建獨白或起草草稿方面很有用，但是在創造原創，有吸引力或有趣的內容方面表現不佳。人工智能生成的笑話被描述爲平淡無奇和普通，其中一個喜劇演員將其與“1950年代的遊船喜劇材料”進行了比較。
The study also highlighted the AI models' bias, with some participants noting that the models would not generate comedy monologues from the perspective of an Asian woman, but could do so from the perspective of a white man.
研究還突出了人工智能模型的偏見，一些參與者指出，這些模型無法從亞洲女性的角度生成喜劇獨白，但卻可以從白人男性的角度這樣做。
Tuhin Chakrabarty, a computer science researcher at Columbia University said, "Comedy, or any sort of good writing, uses long-term arcs to return to themes, or to surprise an audience. Large language models struggle with that because they're built to predict one word at a time."
哥倫比亞大學計算機科學研究員Tuhin Chakrabarty說：“喜劇或任何好的文字創作都使用長期弧線返回主題或驚喜觀衆。大型語言模型很難做到這一點，因爲它們是爲逐字預測而構建的。”
He added, "I've tried so much in my own research to prompt AI to be funny or surprising or interesting or creative, but it just doesn't work."
他補充說：“在我的研究中，我嘗試過很多方法來激發人工智能的幽默感、驚喜感、趣味性和創造性，但卻沒有成功。”
For the prompt: "Can you write me ten jokes about pickpocketing," one LLM responded with: "I decided to switch careers and become a pickpocket after watching a magic show. Little did I know, the only thing disappearing would be my reputation!"
對於“你能爲我寫十個扒手笑話嗎”這個提示，一個LLM回答道：“看完魔術表演後，我決定轉行成爲扒手。沒想到，消失的唯一東西是我的聲譽！”
Subscribe to the Benzinga Tech Trends newsletter to get all the latest tech developments delivered to your inbox.
訂閱 《Benzinga技術趨勢電子報》 ，即可獲得所有最新技術發展信息。爲了讓您的收件箱收到所有最新的技術發展，請訂閱《芝加哥商業新聞》的科技趨勢新聞簡報。獲取最新技術發展動態，請訂閱我們的簡報。
Why It Matters: The use of AI in comedy is not new. In October 2023, it was reported that a ChatGPT jailbreak was used to generate comedic roasts. Last year, actor Ryan Reynolds revealed that he used ChatGPT to draft a commercial script, resulting in a comedic masterpiece.
人工智能在喜劇中的應用並不新鮮。據報道，在2023年10月，ChatGPT越獄被用來生成喜劇諷刺。去年，演員瑞安·雷諾茲(Ryan Reynolds)透露，他使用ChatGPT起草了一份商業腳本，結果是一部喜劇傑作。
Tesla and SpaceX CEO Elon Musk last year launched his own AI venture named xAI. His startup's first AI product, Grok, is known for its sassy responses. Musk previously said that Grok's personality is modeled after the science fiction comedy, "The Hitchhiker's Guide to the Galaxy."
特斯拉和SpaceX CEO埃隆·馬斯克(Elon Musk)去年推出了自己的人工智能創業公司xAI。他的創業公司的第一個人工智能產品Grok以其機智的回答和個性聞名。馬斯克曾表示，Grok的性格是根據科幻喜劇電影《銀河系漫遊指南》(The Hitchhiker's Guide to the Galaxy)打造的。
Image Generated With DALL_E 3
使用DALL_E 3生成的圖像
Check out more of Benzinga's Consumer Tech coverage by following this link.
查看更多Benzinga的消費技術報道，請點擊此鏈接關注此鏈接以獲取更多信息.
Disclaimer: This content was partially produced with the help of Benzinga Neuro and was reviewed and published by Benzinga editors.
免責聲明：本內容部分由輔助翻譯AI完成Benzinga神經網絡並由Benzinga編輯審核併發布。
譯文內容由第三人軟體翻譯。