Human and algorithms are complementary. Now in order to be complementary, one must complete the other weakness.

Examples of what are algorithms good at:
– optimizing schedule with known constraints
– identifying weak signals within a large databases
– being rigorous and data driven only

Examples of what humans are good at:
– Ensuring coherence of a result
– Giving meaning beyond words
– Making sense of weak signals
– Playing with limits
– …

Examples of complementarities:
– Have an AI restricting potential health issues among the realm of possibilities, so the doctor can decide faster with even possibilities thst he would have not thought about.
– Have an AI preparing a compilation of articles on a specific topic analyzing data from multiple sources; have an Analyst preparing an engaging paper out of it.
– Have an AI filtering calls and answering generic questions; have a subject matter expert answering specific questions.
– Generally have an AI doing thing that are repeatable, specific and supervisable; saving time for the human being to add real value to the draft.

AI is not yet smart compared to a kid. However when you are properly restricting the tasks knowing the current limits; then you create new capacities for AI.