Tech Distilled: What F1 Racing Taught Me About AI Data Curation

I’ve never been a motorsports fan. Then, I discovered Drive to Survive on Netflix, and I have to admit—it’s good. It’s like real reality TV, but I do admit it’s dramatized. Yet, it has what every good story needs: drama, politics, character development, suspense (the whole bit with Grosjean and his crash!) all in a neatly wrapped, entertaining bundle.

Before watching it, I didn’t fully appreciate just how technical Formula One racing is. Teams are given a set of rules and regulations within which to operate, and they go off and build the best car they can, min/maxing every possible parameter to gain an advantage. I think that is what I find so fascinating about it.

Engineers Need Constraints to Innovate

I was deep down the F1 rabbit hole, when a fellow colleague recommended me a video of Neil deGrasse Tyson explaining the physics of F1. Part way through the clip, Tyson says something that really caught my attention:

“Engineers love constraints. If you say, ‘Do it with anything you need!’ they’re lost.”

Ask an engineer to just build a car, and the first thing they’ll do is ask you questions: What purpose does this car serve? Will it carry passengers or cargo? Is it a street car, or an off-road car? They are seeking constraints; searching for rules; trying to hone in on a definition of success.

That’s when it occurred to me: data curated for AI requires constraints in the same way that engineers do.

Why Human Structure Fails AI Systems

Here’s where things got interesting for me.

When organizations look at their unstructured file data landscape, they see a type of structure – regional servers, departmental shares, folders, and business-unit repositories. To humans, this looks organized, as people know where things live. If gaps appear, tribal knowledge fills them in.

But AI doesn’t always see things the way people do.

While these structures provide physical separation and can be navigated by humans, it’s not the kind of constraint that AI recognizes—or respects. In fact, it can produce the opposite with AI: duplication, drift, and subtle inconsistencies between copies of what a human may think of as the same data. The result? Hallucinations, something we have all experienced with AI.

For artificial intelligence to be powerful, we have to stop applying the kind of constraints that work for human intelligence. AI demands more precision. It requires full and appropriate context of the data source. It needs all the information necessary for the model to make a correct, relevant, and safe decision, but not be overwhelmed by noise. Without it, AI answers become inconsistent. It looks like the AI solution failed when, in fact, it was an upstream engineering failure.

Consolidation is the Foundation of AI Performance

In F1 terms, imagine if each engineering team designed their systems without a shared specification. Suspension has one goal. Aerodynamics another. Power-units a third. Each team is operating within their own constraints, but they are local, not global. The final machine would be unstable and unreliable.

The same thing happens when different departments manage data independently – each team creates a structure that works for them. Then, when AI operates across the entire environment, those localized rulebooks conflict.

This is why consolidation has to be the first step in the AI journey. Without it, curation is almost impossible. Only when dispersed data is consolidated and brought into a unified and authoritative foundation, can organizations begin to define:

• What is authoritative
• What is current
• What is historical
• Who has access

These definitions are your equivalent of the F1 rulebook, the global boundaries that all business units can optimize around.

Turning Constraints into Competitive Advantage

Lewis Hamilton and Mercedes didn’t win all those titles because they had limitless freedom. They won thanks to disciplined engineering within clearly defined boundaries. AI works the same way. If unstructured data is scattered across silos, AI can only optimize inside each local structure. However, with consolidation and curation – with unified global rulebooks – AI can perform with precision and purpose. It is the constraints that allow AI to move up into pole position.

Tech: Distilled gets to the heart of today’s file data challenges without the fluff. In this series, Ryan Miller, Senior Solutions Architect at Nasuni, unpacks complex technical concepts with sharp insight and real-world relevance. From data security and file locking to the building blocks of a unified file data platform, it’s the kind of practical knowledge that sticks. If you want your tech smart, clear, and just a little bold, this series is for you.