AI Projects And Vintage Cars
As we look back at the history of transportation, we can see how the first cars looked similar to horse-drawn carriages because of the limitations of the time. Early car designers were restricted by the technological barriers of the era, so they used familiar materials and techniques to build new inventions. Fast forward to today, and we can see how a new kind of automation is beginning to take hold, driven by artificial intelligence (AI).
Just as cars replaced horses via automation, AI is beginning to replace data-based decision-making, also known as business intelligence (BI), through automation. But just as the first cars resembled horse-drawn carriages, the modern AI data stack resembles legacy BI data stacks, and many of those similarities are unnecessary or even counterproductive. A data pipeline designed for human review of a dozen dashboard summary metrics is unlikely to be optimal for an AI system that uses hundreds of inputs to make millions of unsupervised decisions about business operations and customers.
To tackle the challenges that come with AI and data management, we need to design a new modern data stack specifically for AI. This new stack will need to address challenges such as governance, data volumes, compute requirements, and time to market, in order to deliver on the full promise of AI in modern business operations.
The modern AI data stack has evolved significantly for automated machine learning (auto-ML) and machine learning operations (MLOps). However, the same cannot be said for the current generation of data pipeline tools and practices. These tools and practices still borrow too much from legacy business intelligence (BI) architectures and practices, which are not necessarily optimal for AI.
To truly take advantage of the full potential of AI, we need to recognize the need for a new and distinct approach, just as cars replaced horses for transportation. It’s time for a new generation of data pipeline tools and practices that are specifically designed for AI.