RDBMS were not as efficient as using AI.
With an RDBMS the operator writes a query they are asking a question and only receives an answer to that single question. It is down to the imagination of the initial operator as to how informative the system ends up being. Most just set up a string of queries and then have the reports generated, daily, weekly, or monthly.
Using AI could throw out all sorts of previously unknown patterns that you would have no clue as to what the fcuk they meant until you sat down and analysed them. My premise of data mining a data warehouse was to have surprises pop up so that people would literally say, "I didn't know that", and then use the information for competitive advantage.
I have not done anything in the field for over 20 years. I do chat and meet people who are still active in the field and it is interesting to hear what advances have been made and how much stuff has advanced. Shopping centres with facial recognition, gait analysis, the ability to lock onto customers and track their route through the shopping centre. To the extent that AI based visual image analysis can identify suspect behaviour in potential thieves, and S vest wearers.