Not the first, and won't be the last to make the point. If AI were making a true impact productivity would be soaring and job numbers dwindling. Neither are. 

Plus most supposed AI projects are minor automation processes. I've witnessed these- AI with IBM Watson doing no more than pulling disparate data sources together.

Take insurance and the use of AI to detect and prevent fraud. In fact you can reduce such fraud by 30% plus just by introducing self-service processes.

  1. Customised claims forms in which all data is analysable including free form text
  2. Claims handlers send links to claimant asking them to take photo's & video of damage to homes, contents or vehicles
  3. 30% walk-away by those taking a chance the insurer will not realise they are conflating old and new damage
  4. Substantial cost reduction as less claims to process and more time to look after bona fide customers
  5. Then access all the data in the enterprise to spot potential fraud- it is alarming the amount of data enterprises cannot access & analyse- more than 70% usually
  6. Once you can access all internal and relevant external data start thinking about AI. You'll be able to afford it and have the time to plan and organise properly.

AI will have it's day but for most enterprises it is some day off