Using machine learning (ML) and then artificial intelligence (AI) to automate business processes is a hot topic and on the wish list at most organizations. However, many, including yours truly, warn that automating business processes based on data with data quality issues is a risky thing.
In my eyes we need to take a phased approach and double use ML and AI to ensure the right business outcomes from AI automated business processes. ML and AI can be used to rationalize data and overcome data quality issues as exemplified in the post The Art in Data Matching.
Instead of applying ML and AI using a dirty dataset at hand for a given business process, the right way will be to use ML and AI to understand and asses relevant datasets within the organization and then use thereon rationalized data to be understood my machines and used for sustainable automation of business processes.
Most of these rationalized data will be master data, where there is a movement to include ML and AI in Master Data Management solutions by forward looking vendors as examined in the post Artificial Intelligence (AI) and Master Data Management (MDM).
Agree. Quite a lot of the solutions from the 2020s (AI, ML, RPA etc.) will not solve any of the problems from ca. 1995 (lack of governance, lack of fundamental overview and understanding of data, lack of understanding that data governance is not a one-off project, organisational setup that is unsuitable and unequipped for managing data at scale etc.)
Unfortunately, getting budget and buy-in for a solution from the future tends to be much easier than getting the same commitment to solving the problems from the past – if you don’t understand the relationship between them 😦
But OTOH if all that basic stuff worked perfectly what were we supposed to spend our time on? 😉
Indeed – automating data governance is still quite futuristic, so we still need to spend (more) time there before we will see business process automation flourish.