Why do enterprises buy into artificial intelligence systems? Where are they investing the most? As you go to your C-suites and boards with new concepts, where should you direct your pitches? Business process automation and customer support are foremost on the minds of executives and managers buying or implementing AI systems, and where many of the budget dollars.
That's the word coming out of a survey of 100 executives by Leverton, which looked at corporate motivations for AI acquisitions. The leading categories of use cases seeing AI investments and work include the following:
- Business process automation 49%
- Customer support/Chatbots 47%
- Data extraction 43%
- Contract analytics 28%
- Voice/video processing/imaging 25%
Business process automation and customer support appear to be the low-hanging fruit when it comes to initial AI rollouts -- and a good place to start for nascent AI programs. "The most prevalent use cases for AI are those for which solutions have had time to mature, the problems solved are less complex and manual precedents are established," the survey report's authors observe.
Still, data extraction and analytics are emerging as AI use cases as well as of late. These use cases "are vastly more complex, especially when dealing with unstructured documents," the Leverton analysts state. "These documents may include thousands of different versions of the same question, all with variations in grammar, syntax, language and layout. The AI needs millions of records before it can begin to be effective. As a result, current solutions are in earlier stages of development, and implementation is a longer and more
In these early stages, AI is not seeing a huge share of technology budgets. Overall, 37 percent projected they would invest less than $250,000 in AI over the coming year. On the high end of the spectrum, only 15 percent anticipate they will sink more than $1 million into such efforts. By use, case, business process automation was the function for which the highest percentage of respondents reported expecting to invest more than $1 million, cited by about 20 percent. "Since this is the area most directly correlated to back office efficiencies, and ultimately ROI, it makes sense that organizations would allot larger portions of their IT budgets to this technology."
Executives were asked how the reality of AI implementations meshed with their expectations. The area that surprised them most was the level of involvement needed for the AI implementation, cited by 30 percent. A similar percentage cited the complexity of their AI efforts as more than initially expected.
By use case, voice, video and image processing was the most difficult to implement, in terms of time (46 percent said it took longer than planned), complexity (42 percent), and involvement (38 percent).
Data extraction was the second most oft-cited challenge, with about 35 percent reporting spending more time than expected, and 35 percent requiring greater involvement.
Staffing is another challenge, but changes with AI maturity -- 42 percent of the "laggards" in the survey indicate that they had data scientists in house, compared to 70 percent of the "leaders."