Glossary What is Generative AI?
By null / 1 Jan 2023 / Topics: Data and AI Digital transformation
By null / 1 Jan 2023 / Topics: Data and AI Digital transformation
Generative Artificial Intelligence (AI) is software that uses machine learning models to create information — including text, images and music — that is similar to human-created content. It is generated using user-provided instructions.1 The instructions are often referred to as prompts or inputs, and the content produced by the generative AI is an output. Generative AI outputs are informed by training data — information provided to the model — like general AI tools, such as conversational chatbots that depend on predefined responses.
However, generative AI expands upon AI capabilities by providing unique outputs that weren’t included in the generative AI’s original programming or training data. This technology has grown in popularity since the release of OpenAI technologies, including ChatGPT, a text-based chatbot, and Dall-E, an image generator.
Generative AI is a stepping-stone technology that OpenAI hopes will lead to artificial general intelligence, which is an AI type that can “solve human-level problems”2, meaning that the model would be able to learn as a human or animal does to complete tasks. To build its generative AI, OpenAI uses deep learning, a machine learning methodology that uses many processing layers to create an output.
Generative AI use cases have been aggressively explored by businesses, researchers and individuals since ChatGPT public testing began in November 2022. The wider implications are being assessed on how the tool can be used, but many time-saving tactics and strategies have been discovered and are heavily discussed.
For instance, ChatGPT can quickly create and organize text-based content, which is used to help people write emails and essays or organize large amounts of data. Tools like Dall-E can then be used to create art works, which have won competition awards.3
Users can lean on generative AI tools to help with a variety of roles, from writing to coding and beyond. The tool has been used to create full reports, provide answers to complex questions and serve as a fact checker. But it’s possible for the outputs to be incorrect, making it critical to have a human review and edit the outputs as necessary.
Creating a generative AI tool for business use can be difficult and resource intensive. Many organizations can adopt a generative AI model similar to how they might adopt a third-party chatbot tool.
While using publicly available tools like ChatGPT can be tempting for organizations, there are potential risks to proprietary and confidential data. This is because user inputs into ChatGPT are added to its learning algorithm. For example, if a user inputs confidential data and asks the program to summarize it, the inputs may be added to the program’s data and may be part future outputs served to other users. This makes it important for organizations to use internally controlled versions that can be hosted by cloud platforms through an Application Programming Interface (API).
Before adopting generative AI, it’s important to work with your IT, legal and business departments to create a policy that keeps the business safe from the risk surrounding questions tied to input and output ownership and data privacy. This tool is in its early development and adoption stages, but offers promise for how it can streamline business processes and help organizations achieve their goals.
Citations
1 Lim, W.M., Gunasekara, A., Pallant, J.L.., Pallant, J.I., Pechenkina, E. (2023, July). Generative AI and the Future of Education: Ragnarök or Reformation? A Paradoxical Perspective from Management Educators. The International Journal of Management Education.
2 OpenAI (2023). Pioneering Research on the Path to AGI.
3 Roose, K. (2022, Sept. 2). An A.I.-Generated Picture Won an Art Prize. Artists Aren’t Happy. The New York Times.