Since the release of ChatGPT in November 2022, generative AI (genAI) has become a top priority for enterprise CEOs and boards of directors. For example, according to a PwC report, 84% of CIOs expect to use genAI to support their new business models in 2024. Indeed, there is no doubt that genAI is a truly revolutionary technology. However, it’s also important to remember that this is just one type of AI and is not the best technology to power every use case.
The concept of what counts as AI will change over time. Fifty years ago, a program that played tic-tac-toe would have been considered a form of AI. Not so much today. However, generally speaking, the history of AI falls into three different categories.
- Traditional analysis: For the past 40 years, organizations have been using analytical business intelligence (BI), but as technology has become more sophisticated and advanced, the name has shifted to analytics. Analysis typically looks back to the past to uncover what happened in the past.
- Predictive AI: This technology is forward-looking, analyzing past data to discover predictive patterns and using current data to accurately predict what will happen in the future.
- Generative AI: GenAI analyzes content such as text, images, audio, and video and generates new content according to user specifications.
“We work with a lot of chief data and artificial intelligence officers (CAIOs), and they’re trying to understand how generative AI can be used to support use cases and models,” said Thomas Robinson, Domino’s chief operating officer (COO). “Predictive AI remains a mainstay of model-driven business, and future models are likely to combine predictive and generative AI.”
In fact, predictive and generative AI work together, such as analyzing radiology images to generate reports on preliminary diagnoses or mining stock data to generate reports that are most likely to increase in the near future. There is already a use case for it. For CIOs and CTOs, this means organizations need a common platform for development. complete AI.
Complete AI development and deployment does not treat these types of AI as separate animals, each with its own stack. Admittedly, genAI may need a little more power on some GPUs, and the network may need to be beefed up to improve performance in some areas of the environment. However, it’s different if your organization is running a truly huge meta-scale genAI deployment. With Microsoft, there’s no need to build a new stack from scratch.
Governance and testing processes also don’t need to be completely reinvented. For example, predictive AI-powered mortgage risk models require rigorous testing, validation, and continuous monitoring, similar to genAI’s large-scale language models (LLMs). Again, there are differences, including the well-known problem with genAI’s “hallucinations.” However, in general, the process of managing risk for genAI is similar to that for predictive AI.
Domino’s Enterprise AI platform is trusted by 1 in 5 Fortune 100 companies to manage their AI tools, data, training, and deployment. This platform allows AI and MLOps teams to manage complete AI (predictive and generative) from a single control center. By consolidating MLOps into a single platform, organizations can enable complete AI development, deployment, and management.
Learn how to reap the benefits and manage the risks of your genAI projects with Domino’s free whitepaper on responsible genAI.
