Finding the perfect price for a product used to take entire wings within an organization. Accountants and analysts analyzed every cent, figuring out the exact point where demand generates the optimal price.

But that exact, optimal price point doesn’t necessarily need to be static. With dynamic pricing, much of that corporate effort has turned to software. Dynamic pricing shifts the price tag of a product or service depending on the individual searching for the product, the time of year or even the weather outside. In tools like Uber Technologies’ surge pricing, artificial intelligence (AI) runs the show, pushing up prices based on the expected amount a consumer may be willing to pay.

It’s a small but powerful example of how AI has found a place within the corporate world. Many AI tools are so subtle that few even realize they’re interacting with the technology. But the number of companies using AI jumped by nearly 25% last year, according to McKinsey’s 2019 Global Survey of artificial intelligence. Firms using AI across multiple business functions grew from 21% in 2018 to 30% today.

“The results indicate increases in AI adoption in nearly every industry,” wrote McKinsey Global Institute partner Michael Chui, a co-author of the report.

Still, that leaves many companies just beginning their AI journey. If your firm hasn’t embraced the technology, take note of these important AI trends.

Cutting Costs Offer Early Results
According to the McKinsey survey, companies tend to experience the early benefits of AI through reduced costs. About 44% of companies said they experienced cost savings in the business unit where the AI tool is deployed. That’s not a big surprise, says Svetlana Sicular, an AI analyst at Gartner.

“The easiest, quantifiable result is cost efficiencies,” she added. “If you don’t start with something that’s quantifiable, you don’t know.”

For instance, an asset-heavy industry can use AI to predict when a piece of machinery will experience problems or break down and schedule maintenance that limits failures and delays. It’s also relatively easy for executives to compare the differences in maintenance costs before and after implementing AI.

The good news for those in the early stages of implementation: as businesses incorporate AI into more departments, revenue follows. Companies incorporating a significant amount of AI throughout their organization were three times more likely to experience revenue increases of more than 10%, compared to companies in the early stages of introducing the technology.

AI Governance Remains A Hot Topic
But AI governance can also present some trickier issues for companies. Currently, most companies only use AI on a small scale, maybe within one department of an organization. But they scale it up as the tool proves its worth. At that point, executives should focus on “managing risk,” says Sicular.

The best way to do this is by ensuring they treat the tool for scale during the design process. That includes accounting for biases that are inherent in the individuals developing the technology.

For instance, Sicular points to studies that have found judges provide softer punishment after lunch than before it — an inherent, hidden bias in sentencing, based on a judge’s day. If that bias becomes engrained in an AI tool, then it could result in backlash.

To manage this risk, Gartner suggests that companies ensure that the data used to build their AI tools come from trustworthy sources, that they design the algorithms and data system running the AI to be transparent and that the company has a diverse team working on the project.

Companies Are Limited By Their Ideas
Of course, many aspects of AI are held back not by the lack of technology or talent base, but by the lack of imagination of those incorporating the AI.

Sicular recently stayed at a hotel where she could check out via Amazon.com’s voice command service Alexa. The hotel had provided a simple acknowledgement script for Alexa to recite when she went to check out; Sicular had to go to reception to ensure she received a receipt. It didn’t make the process “easy,” says Sicular.

Many companies currently deploy AI tools in sales and marketing because of a number of tasks that are repetitive, repeatable and boring. This includes popular consumer-facing chatbot interfaces.

But organizations need to firmly grasp why they are implementing AI, says Sicular. She points to Australia’s intellectual property office, IP Australia, which uses a chatbot to handle questions from inventors seeking patents. Since securing patents is a time-consuming process, the chatbot answers questions along the way, conducting approximately 40% of customer interactions, according to reports. It frees up resources and provides those filing patents with faster information.

It’s an example of how AI can work well, when the benefits are clear, both for the organization and the end user.