You may have seen the Burger King commercial in which a food attendant looks into the camera and says, “OK, Google, what’s the Whopper burger?” prompting Google Home devices everywhere within earshot to cite the Wikipedia entry. It can be viewed as advertising at its best, using AI devices to help peddle a product right into consumers’ living rooms (all while hacking airtime costs). But it can also come off as annoying and a true sabotage of consumer privacy.

Regardless of where you stand on the issue, the short Burger King ad revealed that AI is taking hold in every facet of our lives, and after decades of planning for its arrival, it’s finally here. You see it when you call a service center for tech support and end up conversing with a chatbot, when you ask Siri how many ounces are in a quart or when you play poker against a computer.

But despite its market acceptance, a 2017 McKinsey report found that aside from the tech sector, it’s still in the early, experimental stage among most businesses, and few firms have deployed it at scale. In its survey of 3,000 AI-savvy executives, it found that only 20% claimed they currently use any AI.

As the McKinsey report indicates, companies need to start experimenting with AI and get on the learning curve or risk falling behind.

What’s Holding Them Back?

There’s no question that implementing AI is not easy. It’s perhaps the most disruptive technology of all time. Legacy IT infrastructure was not made to support machine learning, deep learning algorithms or new predictive analytics models, so companies wishing to build AI solutions need to first modernize their entire IT infrastructure. Additionally, the computing power required to build these algorithms requires GPU server capacity, which is extremely costly.

On top of these challenges, AI requires lots of training data in order to be smart. Companies need to take a full inventory of all of their data across the entire enterprise and supplement that data with what might be needed to solve specific business problems. They can do this by purchasing data, crowdsourcing that data or using an insights-as-a-service partner.

It’s All About Fear