Two scenarios to consider: 1. Firm creates an AI group to provide innovative solutions to its various business problems. A promising, seemingly well-suited pro

Two scenarios to consider: 1. Firm creates an AI group to provide innovative solutions to its various business problems. A promising, seemingly well-suited problem is chosen, but after a few months of requirements gathering and prototyping, it turns out that the AI methods are not very effective. At this point the AI group and the business partners either double down, preferring to save face and live with a sub-optimal solution, or else the AI group abandon the project and the business partners are back to square one, having wasted months, and will now bear all the repeated overhead costs of restarting the project with another set of engineers. 2. Firm has an engineering group who are familiar with AI techniques but are also prepared to rely on 'old fashioned' approaches to get the job done. A problem is identified that initially seems like a good opportunity to use AI methods, but after a few months of requirement gathering and prototyping it turns out they are not very effective. The engineer working on the project then takes off their 'data scientist' hat and puts on their 'software developer' hat and solves the problem with other methods. Since they have already been working on the problem for a few months, no additional problem scoping is required and the business partners get their problem solved with minimal delay.