[Paper Review] AI-Generated Content (AIGC): A Survey

Jiayang Wu, Wensheng Gan+3 more|arXiv (Cornell University)|Mar 26, 2023

Topic ModelingComputer Science87 citations

TL;DR

This paper surveys AI-generated content (AIGC), defining its scope, capabilities, and the industrial chain, and discusses AI models, modes of generation, applications, challenges, and future directions.

ABSTRACT

To address the challenges of digital intelligence in the digital economy, artificial intelligence-generated content (AIGC) has emerged. AIGC uses artificial intelligence to assist or replace manual content generation by generating content based on user-inputted keywords or requirements. The development of large model algorithms has significantly strengthened the capabilities of AIGC, which makes AIGC products a promising generative tool and adds convenience to our lives. As an upstream technology, AIGC has unlimited potential to support different downstream applications. It is important to analyze AIGC's current capabilities and shortcomings to understand how it can be best utilized in future applications. Therefore, this paper provides an extensive overview of AIGC, covering its definition, essential conditions, cutting-edge capabilities, and advanced features. Moreover, it discusses the benefits of large-scale pre-trained models and the industrial chain of AIGC. Furthermore, the article explores the distinctions between auxiliary generation and automatic generation within AIGC, providing examples of text generation. The paper also examines the potential integration of AIGC with the Metaverse. Lastly, the article highlights existing issues and suggests some future directions for application.

Motivation & Objective

• Define AIGC and its key conditions (data, hardware, algorithms).
• Characterize cutting-edge capabilities and advanced features of AIGC.
• Describe the industrial chain and the role of large pre-trained models.
• Differentiate AI-assisted writing from AI-generated writing and provide examples.
• Explore applications (including Metaverse) and outline challenges and future directions.

Proposed method

• Conduct an extensive literature survey on AIGC definitions, capabilities, and applications.
• Classify content generation into text, images, audio, and video modes with examples.
• Explain the three essential AIGC components: data, hardware, and algorithms, and their interdependencies.
• Trace the evolution of generative algorithms (GANs, transformers, diffusion, etc.) and large-scale pre-trained models.
• Discuss the industrial value chain (data suppliers, algorithmic institutions, hardware, midstream platforms, downstream users).
• Differentiate AI-assisted writing (AIAW) from AI-generated writing (AIGW) and provide illustrative comparisons.

Experimental results

Research questions

• RQ1 What defines AI-generated content (AIGC) and what conditions are necessary for its development?
• RQ2 What are the main capabilities and features contributing to AIGC performance?
• RQ3 How is the AIGC industry value chain structured from upstream to downstream?
• RQ4 How do AI-assisted writing and AI-generated writing differ in practice and outcomes?
• RQ5 What are potential applications and future directions for AIGC, including integration with the Metaverse?

Key findings

• AIGC encompasses text, image, and video generation across PGC, UGC, and AI-generated modes.
• Three cutting-edge capabilities of AIGC are digital twins, intelligent editing, and intelligent creation.
• Large-scale pre-trained models offer better generalization, cost savings, faster training, multi-task support, and continuous optimization.
• There is a practical distinction between AI-assisted writing (AIAW) and AI-generated writing (AIGW), with humans retaining creative control.
• The AIGC industry chain spans data suppliers, algorithmic institutions, hardware developers, midstream platforms, and downstream content platforms.