How to Avoid 4 Common AI Mistakes

With artificial intelligence solutions such as ChatGPT dominating headlines recently, many business leaders are asking how they can use these tools in their own organizations.

That’s the wrong question.

Rather than beginning with technologies, we encourage our customers to start by thinking through the business problems they are hoping to solve. This prevents organizations from jumping headlong into flashy new tech that internal stakeholders might not fully understand, and it also gives IT leaders the chance to pause, catch their breath and prepare their environments to maximize new solutions. 

For artificial intelligence, this means shoring up the organization’s data environment, a process that is too often overlooked. In particular, we sometimes see organizations make these four mistakes in their rush to adopt AI.

When leaders look closely into using AI to achieve their business goals — whether that means reducing costs, improving efficiency or speeding up the decision-making process — they often find that the data they need is siloed in different places throughout their organizations.

Many organizations lack a modern data platform or any structured data governance model. As a result, business and IT leaders often don’t know where their most critical data is located, or they may lack critical context around their data. For instance, an Internet of Things (IoT) sensor may provide otherwise valuable operational data, except that stakeholders aren’t sure how often the data is being pulled and therefore cannot effectively incorporate it into AI workflows.

In short: AI solutions are only as good as the data they incorporate, and data is only as good as an organization’s governance policies and practices. 

As organizations seek to integrate and unify their data — and make it more accessible to support AI workflows — security must be a top priority. This means taking strategic steps to protect data environments from traditional attack types, such as ransomware, that continue to cripple organizations across industries. Also, as AI solutions integrate with IoT technologies, business and IT leaders must protect their organizations against attacks on connected physical assets. This is especially true in fields with extremely sensitive assets, such as the utility sector.

AI is an evolving field, and ethical issues are still emerging. These include concerns about authorship and intellectual property, bias and discrimination, and privacy and surveillance.

Many AI solutions operate with a black box model, meaning that humans have limited visibility into how these technologies are arriving at conclusions. In such cases, even solution designers cannot fully understand how data is combined to make predictions.

As organizations increasingly use AI to accelerate decision-making, they must take steps to ensure that these decisions are not based on biased data.

We’re reaching a point where emerging technology is evolving and coming to market more quickly than companies can incorporate it into their operations. This underscores the need for top talent with adaptable skill sets, but these workers are in such high demand that organizations often struggle to fill data science roles and other positions related to their AI efforts.

As a result, many organizations are turning to trusted partners such as CDW for staff augmentation. By bringing in talent from the outside, organizations can identify and build out high-value use cases that yield an attractive return on investment and set AI initiatives up for long-term success.

Story by Jill Klein, the Head of Emerging Technology and the Internet of Things for CDW.  Jill also serves as chair of the CompTIA IoT Advisory Council and is actively involved in industry research projects targeted at accelerating the adoption of IoT.