Creating an architecture for heuristic automation
First, let's get a definition of heuristic: in the literature, it is referred to as applying a solution to an issue without the aim of being the optimal solution, but sufficient to fix the immediate problem that was discovered. Trial and error would certainly match this definition. The Hungarian mathematician George Pólya used the term in his book, How to Solve It, first published in 1945. He provided some practical ways of solving problems.
One of his principles is commonly used in architecture applying ML: if you don't have a solution, assume that you have a solution and see what it does. Keep the good stuff and analyze the bits that didn't work well. Try the iterated solution again and learn from it. This is the base of heuristic automation. It uses heuristic learning that can be leveraged through AI that is able to recognize and learn from patterns. AI will use algorithms and automation – it constantly...