Getting on the Same Page about AI’s Role and Value in Human Workflows

The current noise surrounding artificial intelligence (AI) can be overwhelming between its hype, its promises, and seeing it show up just about everywhere as companies add “AI” components to their product lines. However, one perspective often overlooked is where AI conceptually fits in the bigger picture of how people in the IT field and beyond process data and information. 

When you take a higher-level view, AI’s immediate value becomes pretty clear, especially when it’s positioned within a hierarchy of elements that support any workflow. Understanding this hierarchy is key to understanding the value AI promises. But it also sheds light on another, increasingly important point: how data management is evolving.

The Hierarchy of Workflow Components

Navigating any human-based decision workflow involves graduating through a hierarchy of interdependent constructs. You start with the simplest level of a concept, then add more layers of complexity.

Take learning how to drive a car, for example. You don’t fire up the engine at 16 and let ‘er rip at 80MPH on the closest highway. You learn the basics of how a car works, graduate into how to operate it within the legal limits of driving (speed limits, etc.), progress to driving on side streets, and eventually highways. With this example, each layer of learning is dependent on the one that precedes it and supports what follows it. This is because the human brain learns best by building new concepts on established foundations.

Let’s apply the theory to technology—or, more specifically, how infrastructure is leveraged for workflows and decisions:

Figure 1: Hierarchy of components.

• Data. This is the ground floor of where everything starts and is the most granular level. For instance, 8505551234 is data as a string of numbers.
• Information. This is data with context applied to provide some interpretation of data… a “smartening up” of things, if you will. In my above example, the above data string’s meaning gets refined when I do this: (850) 555-1234.
• Knowledge. This layer is about synthesizing information into useful conclusions. For example: “Hey, (850) 555-1234 is Customer X’s number.”
• Wisdom. This is an aggregation of time and experience applied to knowledge that can ultimately drive action. “That’s Customer X’s number. It authorized this payment on 10/13, so I will mark this transaction as not fraud.”

As you saw, each layer of the hierarchy was dependent on its preceding layer and fed toward the final layer—applying wisdom to knowledge.

AI’s Place in the Hierarchy + a Very Important Maxim to Consider

Artificial intelligence’s natural place in the above hierarchy will (and should) be the Data and Information layers, because Knowledge and Wisdom are exclusively human constructs. This distinction can be tough to grasp, given how human-like generative AI’s interactions can come across. But, when you look closer at how a large language model (LLM) is built and its associated inference is accomplished, this idea is plausible.

LLMs utilize algorithmic calculations within a neural network. Inference uses the model’s mathematics to respond, drawing only from abstractions of lower-level data. Generative AI does not create new information; it simply enables potential new knowledge by aggregating vastly more information than a human can process.

So, yes, AI can be deceiving in implying creation in its responses, but Knowledge and Wisdom can only be accomplished by the human brain. This may seem a bit specific and pedantic, so here’s an example that supports my point:

Only human brains can build an emotional bridge to understanding why tomatoes would make a terrible addition to the fruit salad you take to your neighbor’s barbecue.

AI Shook the Data and Information Tree

One of the most compelling things about AI’s growth has been the way it’s assigned new dimensions of value to an organization’s data and information. Before “data was the new oil,” data’s value had a one-to-one relationship with its workload. For example, financial-related data aligned with the CFO office and accounting system, customer-related data was leveraged by sales and customer service, and marketing-related data was used for business development and marketing operations.

Generative AI LLMs inference that leverages retrieval-augmented generation has multiplied and spread that value around. Now organizations have the ability to quickly process and analyze siloed data as a single entity at layers and depths that would take a human mind years to accomplish.

Where Does Pure Storage Fit in the Hierarchy?

The value of Pure Storage to the above construct lies in its data management capabilities. Next-level data management is critical to AI model building and RAG inference—and Pure Storage is already ahead of the curve. Our Enterprise Data Cloud (EDC) vision is perfectly aligned to AI’s demands. The EDC delivers an experience that will prove invaluable as AI is increasingly adopted into the enterprise.

The next blog in this series will explain our EDC and platform values with GenAI as a sample concept in depth. Get an incredible breakdown of the EDC from Pure Storage CEO Charlie Giancarlo.

Until then, keep tomatoes out of your fruit salad, even if AI tells you it is OK.

Manage your data, not your storage.

Discover the EDC