Generative AI has immense potential to solve unsolved business problems.
However, its output is probabilistic in nature and needs to be refined with guard rails.
It is also important to select a use-case that can be reviewed and improved by internal users.
There are many Generative AI models one can use. Some are free, downloadable, and for commercial use. Some are available through APIs.
Decision to start with a model depends on your use-cases’ confidentiality, capability, and commercial needs.
Generative AI models need to be trained iteratively with prompts and feedback to achieve an acceptable level of accuracy.
Training also involves defining the don’ts ensuring that its output is not offensive and misleading.