It's true that computer technology is still not very good at human nuances that are important for business, but that technology will improve over time. For now, the technology is very good as assimulating large chunks of data and determining patterns, including information regarding inventory and sales. Given that real time data, a human can make more effective business decisions.
I have to say that I am impressed with some of the technology now used to answer phones, in which a human-sounding voice can actually respond to answers and questions from the customer who is phoning. At least one has the sense that some progress is being made in dealing with a concern, rather than (as in the past) being put on hold for lengthy periods of time.
An artificial intelligence system can never understand the nuances of the problem as a person can. Each business has its own culture. Cultural elements and human behavior are the most difficult for a computer to understand. I think the best idea is to have the computer make suggestions, and then people consider them.
Some information technology solutions lend themselves to human error more than paper-based information methods. This is especially true in the medical field, where people have been shown to make more mistakes on electronic forms than on hard copies. If I had to guess, I'd say that trend would apply to other areas as well.
I'd say the problem with using artificial intelligence to solve business problems is that the technology is not sufficiently developed and there are many business problems that AI would therefore not be able to deal with. There is too much of a human side to business to feel secure trusting machines (at our current level of AI technology) to truly understand business problems.