From chapter 2 we learned why training data has a significant impact on the outputs from an AI model. We also learned that high quality data should:
- Be accurate
- Have sufficient quantity
- Be diverse
- Consider unintended biases
Then in chapter 3 we covered how the type of architecture you chose is related to the type of problem being solved. We discussed that genetic algorithms are efficient for problems with a defined result format, especially optimisation problems. We also developed a basic intuition for how neural networks (NN) function and how there are many types of NN. The NNs are flexible in their purpose and functionality.
In chapter 4 we learned how reinforcement learning and fitness functions are related and how they work and what their purpose is. We also made the distinction between supervised and unsupervised learning.
Finally in chapter 5 we talked about how to best use AI and the considerations that may improve workflow. We outlined how AI performs in tasks like coding, brainstorming, writing copy, and giving feedback, along with its strengths and weaknesses in each category.
Hopefully this understanding is beneficial in your journey with AI, good luck.