As applications get more user data, they can fine-tune their models to:
- improve model quality/performance for their specific problem space
- Decrease model size/costs
- Applications are great at spitting out multiple different ideas to get the creative process going
- Most AI demos are one-and-done: you offer an input, the machine spits out an output, and you can keep it or throw it away and try again. Increasingly, the models are becoming more iterative.
