At some point, you are likely to end up in a situation where someone is trying to sell you their “AI product”. Do not become overwhelmed. Take a deep breath and use these critical questions to start separating marketing fluff from product substance:
What other methods /models/algorithms/techniques did you try before you decided on what you have today?
Why did you choose this over the others?
Why do you think this outperforms the others?
Has someone else solved a similar problem? If yes, which method did they use?
How easily can the Machine Learning model be retrained on new data or replaced with a more effective version?
What is the time required to re-train?
What are the resources required to re-train?
How similar is the training data compared to the ‘real world’ data that the model will be working with?
How much data is enough?
What are the limitations of the system?
What are the strengths of the system?