Dispelling the 3 most common myths about AI and big data

Register now

It's easy to see why organizations are looking to the potential of artificial intelligence to harness big data. From self-driving cars, robotic hotel concierges, and intelligent delivery drones to unlocking industry-specific insights and actions to outperform the competition, the combination of AI and big data can be a remarkable game changer. That’s why companies have doubled down on investing dollars and resources into these types of initiatives.

But what is often overlooked is that AI and more specifically machine learning, can hamper your efforts especially if combined with big data. Why? Because AI is actually not intelligent, in that it still suffers from the garbage-in-garbage-out problem! Adding big data into the mix makes it even more challenging to find and harness critical, accurate and useful insights these companies so desperately seek. This is the kind of information that will give you a leg up on the competition, drive new revenue and build more rewarding customer relationships.

Before you embark on an AI and big data initiative, there are three fundamental myths you should understand.

MYTH 1: More data equals more insights.

When it comes to AI, more data is not necessarily better. Simply accumulating large amounts of data without a strategy to leverage it won’t give you the results you seek. You’ll likely be waiting a long time to realize any value at all. Identifying the valuable insights to meaningfully improve your business performance is like trying to find needles in haystacks. The solution to this problem isn’t to accumulate more hay, it’s to have an effectively strategy to sift through the hay using AI.

An alternate approach can deliver more effective and timely results: quality over quantity, leveraging smaller data sets but with higher relevance to the problem at hand. If you have high-quality data, and it’s tightly focused around your goals, the quantity of that data becomes a lot less important. Smaller more relevant data beats big noisy data any day when it comes to application of machine learning.

MYTH 2: AI is a silver bullet to getting your data honed and well organized.

There is no magic shortcut to proper data preparation, even with AI. To achieve the desired results, you are going to need clean, accurate data. For example, autonomous car companies drive millions of test miles to train their AI models. There is no algorithm that will allow them to skip that part of the process if they want to develop an intelligent and reliable self-driving car.

Or course, enterprise data can be challenging. It originates from disparate systems, incompatible formats, assorted technology initiatives, business reorganizations and mergers and acquisitions. But you need to build a clean core data model and provide a mechanism to make that data available for training and analysis. There's a lot of upfront work involved before you can apply machine learning to your data.

MYTH 3: AI is inherently smart.

While many believe that AI is inherently smart, the fact is that AI algorithms are actually not so smart. They can lead you down the wrong road if you don’t have a clear understanding of what you want to achieve. That's why social media networks try to limit their use of AI when banning offensive content. Without human oversight, the AI algorithms would remove a lot of useful and proper content from their networks.

The intelligence in AI comes from the humans who train it. For instance, consider how Google search results keep improving based on the answer the users pick. Or how Facebook asks you to tag a friend in a picture, then guided by your input, will provide more accurate tag suggestions to you and your friends. We are the humans assisting the AI!

Human-assisted AI can work in a similar way with enterprise data. Humans can determine where the automation is delivering an accurate result, and where it's not. When a higher output accuracy is needed, human expertise can facilitate that. And by doing so, it’s training the whole system to perform better.

Ultimately, the best advice is to understand where the AI hype meets the practicality of AI's limitations. The bottom line is that all AI and big data initiatives need smart human assistance to function successfully. And when you strike that balance correctly, you’ll be on your way to achieving your game-changing results.

For reprint and licensing requests for this article, click here.