由于首席执行官埃隆·马斯克的混乱，推特继续陷入困境，一位用户注意到推文回复有一些奇怪的现象。经过深入研究，用户Pieter levels(在该平台上名为@levelsio)发现，一些用户显然现在正在使用流行的文本生成人工智能ChatGPT来回复推文。
这一切都始于上午8点56分的一条推特，Levels用典型的两行表情包格式开玩笑说他的游牧朋友们。七分钟后，Levels收到了一个名为Pragyan Tripathi的账号的回复，他自称是一名开发人员、教育家和创始人，自称在科技行业工作了十多年。Tripathi回复《Levels》:
观察游牧民族的生活方式随着时间的推移而发生的变化是很有趣的。
无论你是喜欢尝试新的食物和体验，还是更喜欢坚持熟悉的习惯，最重要的是玩得开心，充分利用你的旅行。
这是一条看似无害且连贯的回复，直到Levels说他开始深入挖掘，因为他怀疑这条回复可能不是由人类生成的。“我有一种直觉，但现在我证明了这一点，”Levels在收到Tripathi的回复后不久在推特上写道。“人们现在有使用ChatGPT自动回复人们推特的人工智能机器人。”
根据莱斯记录他工作流程的推文，他指示ChatGPT回复他关于游牧朋友的推文，人工智能吐出了一段与特里帕蒂的回复非常相似的文字。莱斯还表示，他认为特里帕蒂账户上的所有推文都是使用ChatGPT生成的。
特里帕蒂在推特上回应了level最初的说法，证实他的回复是用人工智能生成的。Tripathi写道:“最近一直在玩ChatGPT + Twitter API自动化。”当Gizmodo向特里帕西询问Level关于他的整个账户都是用人工智能生成的指控时，特里帕西否认了这一说法。特里帕蒂无法准确估计他的推文中有多少是使用人工智能生成的。
leveles没有立即回复Gizmodo就此事发表评论的请求。
ChatGPT上个月刚刚向公众开放，已经在互联网上引起了不小的轰动。虽然其他基于文本的人工智能生成器已经存在了一段时间，但ChatGPT不仅能够生成连贯的句子，而且能够将能够逻辑地传达单个观点的信息段落串在一起，这是以前的人工智能所缺乏的。ChatGPT甚至在Gizmodo上写了一篇文章。正如伊森·莫里克在《哈佛商业评论》中所写的那样，“这是一件非常重要的事情”，那些能够利用这一转折点的人将拥有一个重要的工具。
原文：
As Twitter continues to circle the drain fueled by chaos from CEO Elon Musk, one user noticed something peculiar about tweet replies. After digging a little deeper, user Pieter Levels—who goes by @levelsio on the platform—found that some users are apparently now using the popular text-generating AI ChatGPT to reply to tweets.
It all started with a tweet at 8:56 a.m., where Levels jokes about his nomadic friends using a typical two line meme format. Seven minutes later, Levels received a reply from an account with a display name Pragyan Tripathi, who calls himself a developer, educator, and founder, while boasting over ten years in the tech industry. Tripathi replied to Levels:
It’s interesting to see how the lifestyles of nomadic individuals have changed over time.
Whether you enjoy trying new foods and experiences or prefer to stick to familiar routines, the most important thing is to have fun and make the most of your travels.
ADVERTISEMENT
It’s a seemingly innocuous and coherent reply to Levels’ tweet, until Levels says he started to dig a little deeper based on a suspicion that the reply may not have been generated by a human. “I had a gut feeling but now I proved it,” Levels tweeted shortly after receiving the reply from Tripathi. “People now have A.I. bots running that use ChatGPT to automatically reply to people’s tweets.”
According to Levels’ tweets that document his workflow, he instructed ChatGPT to write a reply to his tweet about his nomadic friends, and the AI spat out a piece of text that was very similar to Tripathi’s reply. Levels also said he believed that all of the tweets on Tripathi’s account were generated using ChatGPT.
Tripathi responded to Levels’ initial claim on Twitter, confirming that his reply was generated using AI. “Have been playing around with ChatGPT + Twitter API automation lately,” Tripathi wrote. When Gizmodo asked Tripathi about Level’s accusation that his entire account was generated using AI, Tripathi denied that claim. Tripathi wasn’t able to estimate exactly how many of his tweets were generated using AI.
Levels did not immediately reply to Gizmodo’s request for comment on this story.
ChatGPT opened to the public just last month, and has already caused quite a stir on the Internet. While other text-based AI generator’s have been around for a while, ChatGPT isn’t just able to generate a coherent sentence, but instead has the ability to string together paragraphs of information that can logically convey a single point—something previous AI has lacked. ChatGPT even wrote its own Gizmodo article. As Ethan Mollick wrote in the Harvard Business Review, “[t]his is a very big deal,” and those who are able to capitalize on this tipping point will have a significant tool in their toolbox.