There are two questions that always surface when conducting a consultation on implementing AI in business processes:

a) What tasks will be taken over by AI, and when?
b) How effective is AI at performing these tasks?

The answer to the first question requires a taxonomy (classification) of the tasks to be executed. Ultimately, two factors matter:

1. The complexity of the task.

2. The company's tolerance for task failure.

Think of the earliest AI applications, such as gamification for learning—this places us in the lower-left quadrant. With generative AI like ChatGPT, we have seen rapid advancements that progressively move us upward toward the upper-left quadrant.

However, most AI companies today are still in the category of tackling tasks with high failure tolerance. For AI to reach enterprise adoption, we must build on the evidence from the upper-left quadrant and address simpler problems in the lower-right quadrant—but with greater certainty and validated through POCs (proofs of concept).

As AI continues to improve and performance standards for enterprise-level use are established, the frontier will gradually shift toward the upper-right quadrant. Once the first question is addressed, we move to the more critical second question:
How effective is AI at performing these tasks?

Information sourced from Sangeet Choudary.