Helping AI models to meet the real world

Devavrat Shah, a principal investigator at MIT's Laboratory for Information and Decision Systems, has been working on designing methods for AI models to handle second-by-second decision-making using limited computational resources. He co-founded Ikigai Labs, which developed a foundation model for tabular, time series data that can take input from enterprise data and learn as it goes along. The model is an extension of graphical models used in GPS devices and communication systems, and it provides real-time planning on a large scale. The practical implication for engineers building AI systems is the ability to develop methods that can extract information from data at scale in an effective manner.

⚡ Key Takeaways

• Ikigai Labs' foundation model is based on years of research in Shah's lab and was patented and licensed by MIT to the company.
• The model takes tabular data as its input, which is structured data such as the familiar row-and-column format used in spreadsheets.
• The system provides real-time planning and forecasting for large businesses, such as consumer goods manufacturers and pharmaceutical companies.
• The model can handle interdependent processes and make decisions that have implications over time.
• Ikigai was recently acquired by Celonis, where Shah is now chief scientist.

The development of AI models that can handle second-by-second decision-making using limited computational resources has the potential to improve business operations by digitizing processes and optimizing predictions. This technology can be applied to various industries, such as consumer goods manufacturing and pharmaceuticals, to make better decisions and improve forecasting.

✅ Practical Steps

1. Apply the concepts from this article to your own system design, considering the use of graphical models for generic, tabular data.
2. Explore the potential of using foundation models for tabular, time series data in your own applications.

Want the full story? Read the original article.