关于我们
AIGC即AI Generated Content,利用人工智能技术来生成内容,是继UGC、PGC之后的新型内容生产方式,AI写作、AI绘画、AI作曲、AI剪辑、AI动画、AI交互等都属于AIGC的分支。
结合人工智能的演进沿革,AIGC的发展历程大致可以分为三个阶段:
早期萌芽阶段(1950s-1990s),受限于当时的科技水平,AIGC仅限于小范围实验。1957 年,莱杰伦·希勒和伦纳德·艾萨克森完成历史第一支由计算机创作的弦乐四重奏《伊利亚克组曲》。1966年,约瑟夫·魏岑鲍姆和肯尼斯·科尔比开发了世界第一款可人机对话的机器人Eliza。80年代中期,IBM创造了语音控制打字机Tangora。
沉淀积累阶段(1990s-2010s),AIGC从实验性向实用性逐渐转变。2006年,深度学习算法、图形处理器、张量处理器等都取得了重大突破。2007年,世界第一部完全由人工智能创作的小说《1 The Road》问世。2012年,微软公开展示了一个全自动同声传译系统,可以自动将英文演讲者的内容通过语音识别、语言翻译、语音合成等技术生成中文语音。
快速发展阶段(2010s至今),深度学习模型不断迭代,AIGC突破性发展。2014年,对抗生产网络GAN出现。2021年,CLIP模型出现;OpenAI推出DALL-E,主要应用于文本与图像交互生成内容。2022年,深度学习模型Diffusion扩散化模型的出现。
新模型下的AIGC所向披靡
过去,互联网的内容都是由用户生成、上传,AI只能协助人类完成一部分最简单、最基础的工作,无法独立生成内容,更不用提优质内容了。但这一状况也因Diffusion扩散化模型的开源应用而被打破,AIGC成为了继UGC之后的又一大内容生成方式。
相较于UGC,AIGC的最大不同是新技术驱动了机器智能创作内容,这使得AIGC具有独特的技术特征,包括数据据量化、内容创造力、跨模态融合、认知交互力等,也正是这些独有的技术能力,让AIGC成为“不可替代”的新一代内容生成方式。
让我们期待,未来AIGC给我们带来的更多精彩!
ABOUT AIGC
AIGC refers to AIgenerated Content, which uses artificial intelligence technology to generate content. It is a new way of content production after UGC and PGC. AI writing, AI painting, AI composing, AI editing, AI animation and AI interaction all belong to the branches of AIGC.
Combined with the evolution of artificial intelligence, the development of AIGC can be roughly divided into three stages:
In the early embryonic stage (1950s -- 1990s), AIGC was limited to small-scale experiments due to the scientific and technological level at that time. 1957. Legeren Hiller and Leonard Isaacson complete history's first computer-composed string quartet, the Illillac Suite. In 1966, Joseph Weizenbaum and Kenneth Colby developed the world's first conversational robot, Eliza. In the mid-1980s, IBM created Tangora, a voice-controlled typewriter.
In the stage of precipitation accumulation (1990s -- 2010s), AIGC gradually changed from experimental to practical. In 2006, breakthroughs were made in deep learning algorithms, graphics processors, tensor processors, and more. In 2007, 1 The Road, the world's first novel written entirely by artificial intelligence, was published. In 2012, Microsoft publicly demonstrated a fully automatic simultaneous interpretation system that can automatically generate Chinese speech from English speakers through speech recognition, language translation, speech synthesis and other technologies.
In the rapid development stage (2010s till now), deep learning model has been iterated continuously, and AIGC has made a breakthrough. In 2014, the adversarial production network GAN emerged. In 2021, CLIP model appeared; OpenAI launched DALL-E, which is mainly used for text and image interaction to generate content. In 2022, the deep learning model Diffusion model appeared.
The AIGC under the new model was invincible
In the past, content on the Internet was generated and uploaded by users. AI could only assist human beings to complete some of the simplest and most basic tasks, and could not independently generate content, let alone quality content. But this situation is also broken by the open source application of the Diffusion model, and AIGC becomes another big content generation mode after UGC.
Compared with UGC, the biggest difference of AIGC is that the new technology drives the machine intelligence to create content, which makes AIGC has unique technical characteristics, including data quantization, content creativity, cross-modal fusion, cognitive interaction, etc. It is these unique technical capabilities that make AIGC an "irreplaceable" new generation of content generation.
Let's look forward to more exciting AIGC for us in the future!