Most people understand the analogy of the last mile in business stories. Sometimes it refers to physical delivery processes, sometimes to internet connectivity, and sometimes it is a general concept. I am going to start using the concept of the last 100 feet in artificial intelligence (AI).
It means the same thing as the other uses of the concept – getting the goods or services into the hands of the consumer. In this case, it is delivering the AI predictions into the hands of the person inside a business process.
In AI the roadblocks in the last 100 feet are conceptual not physical. It is not too difficult to tap into APIs or consume the bulk output of an AI model. Those are tasks related to how software systems and databases interact.
Deploying AI models into business processes is called operationalization. It is the part of the project that ties AI models to business operations. That is how I like to think of operationalization. Most technical people think of operationalization as just the process of connecting the output of the AI model to software applications. That is very important and a definite part of operationalization, but it is not the last 100 feet of AI.
The real last 100 feet of AI is about the acceptance of the prediction by the businessperson. The last 100 feet of AI is less about technology and more about people. This step in operationalizing an AI model is about making business decision makers comfortable with the AI process. Without comfort there will be no adoption of the AI model’s predictions.
The last 100 feet of AI requires a team approach. The data science must be explained to the point that the business decision makers understand the process. It is not about making the businesspeople into data scientists. This part of operationalization is about providing an overview education and answering questions. The answers to questions must come from the perspective of a businessperson and not from the data scientist’s perspective.
I have been working a bit recently on some standardized materials that I have taken to a few projects operationalization stage. I am not yet finding a lot of material that cuts across companies and departments though. Everything seems to be a custom creation to fit with the company’s culture and individual business processes.
Maybe in the future I will find a magic set of pages and definitions that all businesspeople can read to understand AI. I am not betting on finding that anytime soon though.