内容跟不上技术翻新的互联网时代

每个时代都有精致、吸睛的电子时髦玩意儿。在1990年代末和2000年代初期，让人们疯狂传阅的是用Flash软件制作的动画程序，还有以笑话、隽永语录和小故事为题材的电子邮件插图。当时这些数码创作的病毒式传播效应（viral effect）和网络效应（network effect），不下于现在的抖音（TikTok）短视频，和人们通过推特（Twitter）和面子书（Facebook）转发的恶搞式贴图。

我觉得，我们只是在内容载体的技术上不断精进，从电子邮件附件 --> 个人网站 --> 部落格 --> 个性化社交媒体网络，一直演变进化到今时今日能够轻而易举传遍网络、蛊惑人心的抖音式短视频。

但是，如果你还收藏1990末和2000年初期所接收的Flash动画电子邮件附件（Flash animation email attachment）、各种JPG格式的图片电子邮件附件（image email attachment），在对比你在面子书（Facebook）、推特（Twitter）、领英（Linkedin）、抖音（TikTok）等平台上看到的各种图像和视频，你会发现我们所接触到的内容素质不仅没有增长，反而更趋单元、浅薄和俗不可耐。

世代交替和网络载体技术的翻新，并没有为我们提供更优质的阅听内容。平台最大化商业利益的演算法（algorithmn），以推送争议性信息流，助推（nudge）网民发帖争辩、相互标签与抹黑的无止境循环，来达到提高网络流量和点击率的目的，极少促进不同观点网民之间的思维激荡。

ChatGPT所带来的内容通膨现象（详情请查阅凯斌兄在其面子书所转载的文章《AI创造的内容通膨时代来临》），只会进一步加剧互联网通俗和同质性内容泛滥、原创内容稀缺、劣币驱良币的恶果。

内容通膨原本指的是互联网上的内容比过往任何时候都还要多，但对资讯受众而言，内容的价值和意义却正在迅速萎缩。

由ChatGPT等生成式人工智能模型所触发的内容通膨现象，指的则是由AI资讯系统生成的内容量呈指数增长，但生成作品的质量和原创性，却并没有获得相应的提升，导致市场出现内容供过于求，造成内容创作者的收入大幅下跌。这现象背后的成因是，基于大型语言模型（Large Language Model）技术的生成式AI是在海量现有文本内容的基础上进行训练的，这导致大量内容类似或完全相同文本作品的涌现。

换句话说，生成式人工智能倾向于复制和重复，而不是创意和创新。这在某种程度上也正好限制了生成式人工智能系统的创意思维和想象力。

内容的同质化和平庸化现象对内容原创者、消费者和传播平台都构成了严峻的挑战。随着网络文本内容的泛滥，创作者越来越难以吸引阅听人的垂注；消费者则面临在海量千篇一律的内容中找到真正有价值、感兴趣之内容的茫然；而平台则需费思量，在维护用户的忠诚度与满意度、拓展平台流量，与防止虚假和误导性信息的传播之间取得平衡。

问题的解决之道，在于在内容的数量和质量之间取得平衡。创作者必须用心制作有原创性和高附加值的内容来吸引阅听人；而平台需要优先推广真正利惠资讯受众，能够留住忠诚用户的高质量内容。

综上所述，数码网络内容的时代在技术和内容交付机制方面取得了显著的进步。然而，内容的质量并没有跟上技术创新的步伐。由ChatGPT等大型语言模型驱动的内容通膨现象导致了平庸和同质化的内容大量涌现，使得创作者、消费者和平台越来越难以在瞬息多变、海量内容纵横交错的混乱生态中求存。解决这个难题需要所有利益相关者共同努力，将质量置于数量之上，积极推广具有创意的原创内容。

文章发表于2023年3月9日的《东方日报》

参考资料/Reference

[1] https://www.thelifestyle-files.com/what-is-content-inflation-and-how-you-can-avoid-it/

[2] https://www.facebook.com/photo/?fbid=9274081735942928&set=a.245104905507368

Translation by ChatGPT:

The era of the Internet, in which content cannot keep up with technological innovation

Every era has its exquisite and eye-catching digital fashions. In the late 1990s and early 2000s, people were amazed by animated programs made with Flash and email attachments featuring jokes, memorable quotes, and meaningful short stories. These digital creations had a viral effect and network effect no less powerful than the short videos on TikTok today, or the spoofs and memes shared on Twitter and Facebook.

Advancements in technology have greatly improved the medium and efficiency of content delivery over time, from email attachments to personal websites, blogs, personalized social media networks, and most recently, the popular TikTok-style short videos that have the ability to reach and influence a vast audience. However, if you compare the flash animations and various JPG format images that you used to receive as email attachments in the late 1990s and early 2000s to the images and videos circulating on Facebook, Twitter, LinkedIn, TikTok, and other platforms today, it becomes clear that the quality of the content has become increasingly formulaic, shallow, and vulgar.

The generational transition and ongoing technological innovation of online content platforms have not provided us with improved content quality. Platform algorithms aim to maximize corporate profits by pushing controversial information feeds to users and perpetuating an endless cycle of posting, arguing, labeling, and smearing among netizens to increase Website traffic and click rates, while rarely promoting intellectual discourse between netizens with different views.

The phenomenon of content inflation brought by ChatGPT will only further exacerbate the proliferation of mediocre and homogeneous content on the Internet, scarcity of original content, and the problem of bad money driving out good. The article "The Era of Content Inflation Created by AI" details this phenomenon.

Content inflation originally referred to the fact that there is more content on the Internet than ever before, but its value and meaning are rapidly decreasing for information audiences.

On the other hand, content inflation brought by Generative AI models such as ChatGPT, refers to the exponential increase in the amount of content generated without any significant improvement in quality or originality. This is a result of Large Language Models being trained on large amounts of existing content, leading to the reproduction of similar or identical content. While the technology has the potential to facilitate the creation of innovative and original content, the current state of AI-generated content is plagued by issues such as lack of creativity, coherence, and relevance.

In other words, Generative AI tends to replicate and repeat the patterns it has learned from the training data rather than being creative and innovative. This constraint also somewhat limits the creativity and imagination of Generative AI systems.

The proliferation of mediocre and homogeneous content poses a significant challenge for content creators, consumers, and platforms alike. With an ever-increasing amount of content available online, it is becoming increasingly difficult for creators to produce content that stands out and captures the attention of their target audience. Consumers, on the other hand, are faced with the daunting task of sifting through a sea of mediocre content to find something that is of genuine interest or value to them. Platforms, meanwhile, are struggling to strike a balance between maintaining user loyalty and satisfaction, increasing platform traffic, and preventing the spread of fake news and misinformation.

The solution to this problem lies in striking a balance between quantity and quality. Content creators must focus on producing original and valuable content that resonates with their audience, while platforms need to prioritize the promotion of high-quality content that can retain loyal consumers.

In conclusion, the era of digital content has seen significant advancements in technology and content delivery mechanisms. However, the quality of content has not kept up with the pace of technological innovation. The phenomenon of content inflation, driven by AI models such as ChatGPT, has led to a proliferation of mediocre and homogeneous content, making it increasingly difficult for creators, consumers, and platforms to navigate the online content landscape. Addressing this issue requires a concerted effort from all stakeholders to prioritize quality over quantity and promote the development of creative and original content.