Have you ever wondered how AI (artificial intelligence) is both a friend and a foe in the IoT (Internet of Things) industry’s cybersecurity efforts? This column is going to address some of the challenges companies face when implementing AI and machine learning projects in the enterprise.

According to IDC, worldwide spending on AI systems will grow to nearly $35.8 billion by the end of this year, and it will more than double to $79.2 billion in 2022. The five industries leading AI system spending between 2018 and 2022, according to IDC, are retail, banking, discrete manufacturing, healthcare, and process manufacturing.

Let’s begin by looking at some insights from Deloitte’s second annual survey which is showing  early adopters are ramping up their AI investments. In fact, 37% of AI leaders say their companies have invested $5 million or more in cognitive technologies.

What does this investment look like? It is launching more initiatives, in part because the barriers to entry—cost and expertise required to start—are continually lowering.

Second, Deloitte says early adopters need to pursue the right mix of talent to capitalize on AI in the enterprise.

The survey suggests companies are generally short on people who have the technical skills, like AI researchers and programmers, as well as big-picture people, like business leaders who can select the best use cases for enterprise AI.

A third takeaway from Deloitte’s latest AI in enterprise survey is that companies should improve risk and change management, including addressing cybersecurity vulnerabilities.

Let’s take a closer look at security. Of the 1,100 participants in this study, 82% said their businesses have experienced a positive financial return on their AI investments.

If early adopters are experiencing positive ROI (return on investment), the word will spread. More companies will learn from what the early adopters have done, and, after a while, AI will just be part of enterprise operations in all industries.

So, how are these companies using AI?

  • 63% of enterprises have adopted machine learning, making it the most popular AI technology category in 2018. Other AI technologies companies are using include deep learning (50%), natural language processing (62%), and computer vision (57%).
  • 59% are using enterprise software with AI “baked in,” giving them the insights and intelligence they need to achieve their AI-related business goals.

These business goals range from enhancing current products, optimizing internal operations, and making better business decisions to freeing workers up to be more creative with their time and energy.

But there are also hurdles. Alegion just released a survey suggesting a whopping 96% of enterprises encounter training data quality and labeling challenges in their AI and machine learning initiatives.

And, despite all the good news, 78% of enterprise organizations in Alegion’s study said their AI/machine learning projects have stalled at some point before deployment.

This study definitely suggests that data issues are causing project hurdles for enterprises, and 81% of respondents also said training AI with data was “more challenging than expected.”

Deloitte also cited data issues as a top hurdle, alongside implementation challenges, integration challenges, the initial cost, and a lack of skills. The top overall concern, though, was cybersecurity.

One in five Deloitte respondents even said that cyber risk concerns made their companies decide not to launch an AI initiative. Some 16% canceled an AI initiative that was already in progress due to cybersecurity concerns.

What exactly are these concerns? There are several AI-related cybersecurity concerns, ranging from vulnerabilities within the AI itself to the ethical risks of using AI.

Other risks include the legal responsibility an enterprise could shoulder based on decisions made by an AI-enabled system, as well as the failure of an AI system in a mission-critical scenario.

The truth is that AI is a relatively new technology, so I’m not surprised that hurdles like the ones I’ve mentioned are cropping up for a loT of businesses.

It’s important to have these hurdle discussions about powerful technologies like AI. At the end of the day, despite the hurdles, AI is definitely something you should be looking at implementing, if you haven’t already. AI can help the enterprise optimize operations, transform the customer experience, and create new products and services. And in the future, everybody’s going to be utilizing it.

Want to tweet about this article? Use hashtags #M2M #IoT #AI #artificialintelligence #machinelearning #5G #bigdata #digitaltransformation #cybersecurity #blockchain #enterprise #manufacturing #IIoT