You have a meaningful business problem to solve, an excellent data science team, PM skills, budget, software tools, and executive sponsorship. Those are all important ingredients for a successful AI project. There is one thing that will help ensure this project is successful and does not become shelf-ware.  Trust.

Having driven a large scale machine learning forecasting project that went into production, the single most important factor in the forecasting project’s success was not the AI/ML models, but rather the environment of trust that was created throughout the project. Without that trust, the project would have been doomed.

In all AI projects, humans are involved.  To create that environment of trust the project should embrace the principle of 

Combining the best of the Human and the Machine.

The project team and stakeholders have to recognize that

  • the machine can see patterns and find relationships that the human eye cannot see
  • humans know details about the business that the machine can never know.

Tangibly, two layers of trust are required:

1. Trust that people’s jobs will not be taken away by the machine

2. Trust and confidence in the AI/ML models

1. Trust that AI will not take people’s jobs

Stakeholders may perceive that a successful project may result in job cuts.  Address this fear head on by reinforcing that

  • AI can help them do their jobs more efficiently
  • AI can give them more time to work on higher value work
  • AI can give them new skills (tools and techniques) to create more value to the organization

2.   Trust and Confidence in the AI/ML Models

It’s been said that a confused mind always say’s no.  If stakeholders don’t have trust in the in underlying AI techniques, the project will be at risk. In this context trust is driven by understanding. Understanding the techniques used. Understanding how the models work conceptually.  To address this, invest in education of stakeholders on the techniques used. This must be explainable and understandable. Invest in effectively communicating the concepts in explainable and understandable terms: 

  • This training can grow into a program to up-level the talent in your organization. As stakeholders begin to understand more about concepts, two things can happen:
    • Certain employees will rise to deeper ongoing learning and thus you can build data sciences talent from within
    • Employees can better envision future solutions knowing what kinds of problems can be solved with AI solutions

Humans know things that machines don’t. And machines do things humans can’t. Embrace this to get the best from humans and the machine. This will require creating an environment of trust which is contribute to the chances of your AI project being successful.