Using artificial intelligence (AI) to support business operations is not fully understood, and the availability of time and skills, as well as return on investment criteria, must be taken into account to justify investments in AI and ensure that the intended benefits are achieved, says market research firm Forrester senior analyst Brandon Purcell.

He believes that AI will eventually significantly disrupt the way organisations win, serve and retain customers, but not in the short term. To reap the intended benefits, it will take massive amounts of data to train AI systems to perform their jobs well enough to replace their human counterparts, he emphasises.

“AI is not a homogeneous set of technologies, and some tasks will take longer to automate than others. Further, despite the goal of AI technology to free humans from some tasks so that they may more effectively focus on others, the process of creating this state holds significant challenges for human designers and engineers,” Purcell explains.

A hurdle to the adoption of AI into mainstream business is the lack of a clear business case. Research and academic communities were the first to develop and deploy AI technologies, and businesses are only now assessing the technology.

“Organisations require a well-defined return on investment to justify investment in AI.”

AI can bring insights automatically to the surface, with banks today already using such technologies to detect anomalies that might indicate fraudulent transactions, he highlights.

Combing through massive datasets will also allow for better data analysis, particularly regarding unstructured data. As storage and processing power advances, AI is gaining traction among businesses and enabling them to generate insights and more effectively engage with customers.

Purcell says AI is uniquely suited to help optimise customer interactions across touchpoints and channels, largely driven by the technology’s ability to process huge amounts of data, which can inform real-time action.

“In the near future, business leaders will be able to blend technologies, such as facial scanning, text analytics, machine learning and natural language generation to better engage with their customers,” says Purcell.

Time and skills are also potential hurdles. AI systems require massive amounts of training data to learn to perform specific tasks. While some vendors offer solutions, even these will require many additional hours of training and refinement before they can be deployed.

There is a clear dearth of AI skills. There are a handful of notable researchers in academia who specialise in deep learning and AI, but the talent pool for businesses is extremely shallow. Additionally, since AI adoption for businesses is nascent, there are few people with the ability to deploy AI in a business context, he emphasises.

“AI will not be a threat to most jobs. While there may be some losses in the call centre and other positions, for the most part, AI will free employees from tasks that add little value. However, owing to the relative immaturity of AI, we are still in the early stages of its broad use,” he concludes.