Generative AI: Overview

Generative AI, or Gen AI, produces high quality content—such as text, images, video, audio or software code—in response to a user’s prompt or request.
The arrival of ChatGPT in 2022 has thrusted Gen AI into worldwide headlines and launched an unprecedented surge of AI innovation and adoption.
Generative AI is one kind of supervised machine learning. It begins with a foundation model — a deep learning model that serves as the basis for multiple different types of generative AI applications. The most common foundation models today are large language models (LLMs).
For the most part, generative AI operates in three phases:
- Training, to create a foundation model that can serve as the basis of multiple gen AI applications.
- Tuning, to tailor the foundation model to a specific gen AI application.
- Generation, evaluation and retuning, to assess the gen AI application’s output and continually improve its quality and accuracy.
Generative AI offers enormous productivity benefits for individuals and organizations. So far there are two primary categories:
- Web-based interface applications, for instance, ChatGPT, or Bing Chat. Many of use ChatGPT for writing assistance, brainstorming ideas and so on.
- Using generative AI in domain-specific applications: LLMs are used as a development or reasoning tool. This could be an exciting area for supply chain management.