*Machine learning is best used when*: 1. The system you're building can learn patterns from data 2. The patterns in the data are practically too complex to code it in 3. The patterns in the data constantly change, even when they're simple enough to hardcode 4. The data exists or it can be gathered on the way to solving the problem 5. The unseen data shares patterns with the seen data 6. The process is repetitive 7. The cost of wrong prediction doesn't outweigh the benefits 8. It's at scale **Machine learning shouldn't be used when**: 1. It's unethical to 2. Simpler solutions do the trick 3. It's not cost-effective --- noted from the book [[Designing Machine Learning Systems]] by [[Chip Huyen]] also see [[Should we use Artificial Intelligence]]