It’s time to walk the walk with AI | Tech Radar

Overview

When you purchase through links on our site, we may earn an affiliate commission. Here’s how it works.

Image credit: geralt on Pixabay (Image credit: Pixabay)

Details

With all of the ‘hype’, AI can sometimes feel like old news, with nearly every organization discussing how it is impacting their business. But right now, much of the talk doesn’t match up with reality. Most companies wouldn’t want to admit this, but they remain stuck in pilot mode.

In fact, research from MIT’s NANDA initiative found that 95% of AI pilot programs fail, delivering little to no measurable impact. Rather than leaping ahead, most are scrambling to show value, lacking the confidence needed to truly innovate.

Vice President of Product Strategy at Veeam Software.

At the heart of this challenge is data. The sheer scale, complexity, and sensitivity of what’s needed for AI can be intimidating and even paralyzing. And understandably so. Accessing, managing, and securing data in an AI-driven world is daunting, and existing resilience measures often feel inadequate.

Championing data leadership: how can data strategy shape AI success?

AI blindness is costing your business: how to build trust in the data powering AI

The visibility mirage: Why AI pilots keep stalling between ambition and impact

Yet good data hygiene remains essential, and investing in visibility and resilience from the start is the only way to move forward with confidence. Otherwise, you’ll be stuck talking the talk, rather than walking the walk.

With so much of the conversation around AI focusing on the potential for business transformation, it’s easy to forget what it all boils down to - data. Generative AI (Gen AI), Large Language Models (LLMs), anomaly detection, prediction models, you name it, they are all built on, trained on, and create data.

It’s a large part of the reason why we’re expected to create, capture, copy, and consume 181 zettabytes of data globally this year alone, 3 times as much as we did 5 years ago. It’s hard to conceptualize such large numbers, but essentially, there is far more data that organizations were previously equipped to handle.

AI is also flipping the script on how much data companies can actually use. According to Gartner, 80% of enterprise data is unstructured. Before AI, that meant it was mostly just sitting there, often needing to be stored and protected, but impossible to extract value out of it. With AI, that’s all changed.

And, it’s growing exponentially as AI continues to evolve. So today, the real reality of AI is that organizations are struggling to wade through the growing mountains of data to categorize what they’ve actually got. Add an AI pilot program to the top of this pile and it becomes obvious why so many pilots currently fail.

So while organizations would like to say that they’ve got a watertight AI policy in place, for most, shadow IT remains a very real issue. Failing pilot programs are holding organizations back, leaving employees to experiment in the background with unauthorized AI tools.

And this will only continue, unless organizations can escape the sinkhole of data to drive real AI innovation.

How to take AI from pilots to deliver real business value

AI governance under strain: what modern platforms mean for data privacy

AI might be heralded as a ‘new era’, but believe it or not, this ‘new era’ needs to be built on the foundations of the last one. Quite simply, good data hygiene remains good data hygiene, and there’s no need to write off your existing resilience measures.

That means continuing to carry out impact assessments on all of your data. Because the first step to sort through the ever-growing piles of data is to understand what you’ve got. Only then can you identify what data is actually the most integral to your organization, and treat it appropriately.

Gaining that visibility is essential for ensuring the resilience of your data as it continues to grow. Otherwise, if an incident does occur, you won’t know what data you actually need to get back up and running, and you won’t be able to identify the last known good state.

This can’t be a ‘one and done’. The flow of data is not stopping anytime soon, and you need to keep a handle on all of it. It’s integral that practices like data standardization, data validation, and continuous impact assessments continue to stop organizations from getting buried under the flow again.

Hopefully, most organizations should have these measures in place already, so it’s less about bringing in brand new methods to unlock AI and more about expanding what you’ve already got in place. And getting these foundations right really does need to be the first priority.

Because the truth is, AI can actually help you do this groundwork. You can use AI to support data classification, improve your data lineage, and strengthen your resilience measures. In a way, your first AI project should simply be to look after your data. Get AI to look after your data, and then your data will look after your AI.

Establishing this sense of control over your data is essential to building not just the foundations, but the confidence needed to truly innovate with AI, and to deliver pilot programs that actually work.

It sounds obvious, but the solution is simple in principle: start small. You don’t need to invent the next great big thing; you just need to prove that your organization can deliver innovation and drive value while balancing control.

Rather than reinventing the wheel, begin with a manageable initiative where AI can safely add value and demonstrate results. With that under your belt, you won’t just build your own confidence, you’ll also prove to the wider organization that innovation is possible. Then, you can move on to bigger, more transformative applications.

But throughout it all, keep returning to the basics. Ensure that the cost of creation, the performance, and the resiliency of your AI model are all aligned. Otherwise, you won’t be able to build business processes around it without jeopardizing your resilience.

At every step, you should be able to explain the processes, and the second you can’t, is when you need to stop and roll it back before it gets out of control.

Starting small is key to overcoming that fear of failure that is holding so many organizations back from not only harnessing the true value of their data, but also using AI to deliver real, transformative business value.

But we need to carry a small dose of that fear throughout the process to keep that crucial balance between control and innovation to stay resilient. That’s how you finally talk the talk and walk the walk when it comes to AI.

This article was produced as part of Tech Radar Pro's Expert Insights channel where we feature the best and brightest minds in the technology industry today. The views expressed here are those of the author and are not necessarily those of Tech Radar Pro or Future plc. If you are interested in contributing find out more here: https://www.techradar.com/news/submit-your-story-to-techradar-pro

You must confirm your public display name before commenting

1'We've had rough discussions': Virgin River season 8 is still in the early stages of planning, confirms Netflix star

2 Chinese government cracks down on in-office Open Claw use over potential security risks

4'Data is so ubiquitous nowadays, but we’re not necessarily coming up with new ways to protect it': Researchers develop ‘negative light’ signals that hide transmissions in plain sight

5 How to watch Leicester Tigers vs Exeter Chiefs: live stream the PREM Rugby Cup final 2025/26, TV channels

Tech Radar is part of Future US Inc, an international media group and leading digital publisher. Visit our corporate site.

Key Takeaways

When you purchase through links on our site, we may earn an affiliate commission
Image credit: geralt on Pixabay (Image credit: Pixabay)
With all of the ‘hype’, AI can sometimes feel like old news, with nearly every organization discussing how it is impacting their business
In fact, research from MIT’s NANDA initiative found that 95% of AI pilot programs fail, delivering little to no measurable impact
Vice President of Product Strategy at Veeam Software