Executives struggle with deciding which AI work is feasible and which project should be prioritized over competing work. Here’s my list of 7 questions that executives should ask. The quality of these answer tell you a lot about the homework that the team has done.
- Do we have a clear intention and problem statement, with clearly stated inputs/outputs and business goal that define success?
- What is the opportunity cost: AI vs Non-AI vs BAU (not every problem is AI compatible)? what is the risk vs reward-matrix and the expected costs for the entire lifecycle (not just model development and deployment)?
- Do we have ample, high quality, representative data? What is the data provenance? is this in line with our digital maturity? does our data architecture support this initiative?
- Do we have a diverse team (ideally, the AI governance team but not every organization has an official oversight board) who has vetted this for ethical and user experience considerations?
- Was there a ‘Design Thinking’ session in which we considered our assumptions, unknowns, layers of effects (unintended consequences)?
- What’s the governance plan, pre-launch and post-launch?
- How robust is the approach? What happens if things go wrong — do we have resources to test the model against attacks? are there safeguards in place?
Note that I did not include anything about the type of model, devops or deeply technical because as an executive you should delegate those to your experts.