If 2022 can be considered as the genesis of AIGC, and 2023 as its period of accelerated growth, then 2024 will herald an explosive phase of development. Following the waves of PGC and UGC, AIGC as a novel form of content creation has gone through four exponential iterations in less than two years: the first wave marked the emergence of large models represented by GPT-4; the second wave brought rapid innovation in application layer, such as Microsoft's Copilot, which shifted intelligence from chat-based interactions to workflow integration; the third wave witnessed the rise of AI terminals, with AIpin and AIPC taking the stage to empower various industries. The fourth wave saw the emergence of multimodal capabilities, exemplified by Google's Gemini and Pika1.0. Today, AIGC finds wide application in fields like intelligent manufacturing, technological innovation, healthcare, entertainment, and culture. While it brings profound empowerment, it is also entangled in increasingly intense controversies and criticisms regarding its application.
The Catalyst for Efficiency and Cost Reduction
Taking the gaming industry as an example, AIGC assists experienced creators in capturing inspiration and innovating interactive forms. By generating numerous game characters, scenes, and even script narratives based on demand, AIGC significantly reduces the workload and trial-and-error costs for game creators. Furthermore, AIGC has become an auxiliary tool for creative professionals, assisting in ideation and the presentation of innovative formats. It is particularly applicable to industries such as advertising, media, art, and film, enabling the construction of faster market increments with lower costs and higher efficiency.
Emergence of Clear Commercial Models in AIGC Landscape
One of the reasons behind the swift breakthrough of AIGC lies in its potential to become a new avenue for the commercialization of large models on a global scale. While large models have demonstrated promising effectiveness in various fields, their true value as commercial endeavors cannot be realized unless they find widespread practical applications. AIGC possesses the potential to transform business models by automating tasks, improving efficiency, and enabling new operating methods. It no longer needs to be bundled with hardware or systems for commercialization, as it has in the past. Additionally, for ordinary individuals, AIGC is no longer an elusive cutting-edge technology but rather a user-friendly tool that enhances efficiency. This means that the commercial model of AIGC is becoming more evident.
Intellectual property concerns weigh on creators
As the applications of AIGC become more widespread, the voices questioning, condemning, and opposing it have never ceased. The major point of contention lies in the challenges AIGC's commercialization poses to creators, as well as the intellectual property concerns it raises. As AIGC is not a recognized legal entity, it cannot hold copyright as its own subject. However, different views exist regarding copyright ownership of the images generated by AIGC: whether they belong to the platforms, are entirely open-source, or to the creators themselves. Many artists and creators have already declared a ban on AI learning from their works to protect their intellectual property. Websites such as Getty Images and Newgrounds have also announced prohibitions on uploading and selling AIGC works.
A Significant Gap Remains Before Achieving General AI
From a technical standpoint, although the images and text generated by AIGC can already be used for commercial purposes, there are still some challenges that prevent meeting higher quality requirements. It can be observed that AIGC performs well in generating two-dimensional or abstract images. However, when it comes to more specific and detailed content, the generated results may lack satisfaction.
To analyze the reasons behind this, it is evident from the working principles of AIGC that there are still certain errors in its understanding and cognition of natural semantics. For instance, in the case of text generating images, when you input text content about a cat riding an airplane, the resulting image may show a cat sitting on top of an airplane. The quality, compliance, and stylistic factors also influence the generated results. Therefore, to effectively utilize the content generated by AIGC in a commercial context, practical challenges in natural language processing, translation models, and generation algorithms need to be overcome.
