Engineering+Science, and why it is better than Engineering vs. Science debate

Best practices for Artificial Intelligence, Machine Learning, Deep Learning

When it comes to Artificial Intelligence, Machine Learning, and Deep Learning, there are certain best practices that can help ensure a good, usable model. These best practices are based on three key areas: Mathematics, Engineering, and Human factors.

From a human perspective, a good model should be understandable, and the user should be able to explain why the model made the decisions it did. This is particularly important in Deep Learning, where there is a trend of focusing only on the model's accuracy, without considering how the model will perform on new data. It's important to avoid this approach and always use critical thinking.

From a mathematical standpoint, a good model should be generalizable, meaning that it performs well on new, unseen data, not just on the data it was trained on. It's important to avoid over-fitting and over-training, as these practices can lead to models that do well on the training data but perform poorly on new data.

From an engineering perspective, a good model should be repeatable and learnable. Experimentation should be reproducible, and metrics should be tracked so that the next experiment can be improved upon. In other words, learn from the experiment, not just the statistics.

Overall, following these best practices can help ensure that an AI model is effective, usable, and reliable. It's important to remember that AI is a tool, and that a good tool is one that can be understood, is reliable, and is adaptable to new situations.