Diagram
A decision tree starting from the use case that checks whether a pre-trained traditional AI model meets requirements, whether enough training data exists for custom training, and whether control or faster go-to-market is the priority, ending in a pre-trained model, custom-trained traditional model, or generative AI.
Model Selection Decision Flow Under Technical Constraints
The model selection decision flow applies data, control, and speed constraints in order, starting from the use case. The flow first checks pre-trained models, then checks data for custom training, then considers generative AI for speed and flexibility:
| Option | When to choose | Data requirement | Tradeoff |
|---|---|---|---|
| Pre-trained traditional AI model | Classification or detection use case and the pre-trained model meets requirements | No custom training data needed | Fastest path, but less control over model behavior |
| Custom-trained traditional AI model | Pre-trained model does not meet requirements and high control is the priority | Enough training data to custom train | More control over algorithm, loss functions, explainability, layers, learning rate, and hyperparameters; slower GTM |
| Generative AI with prompt engineering | No training dataset exists, datasets are too small to custom train, or faster GTM is the priority | Can start with prompts; can be tuned with data examples | Faster GTM and multimodal support; less control than custom training |
| Traditional predictive AI | Predictive use case with structured data | Particularly effective with structured data | Best fit for prediction with structured data |
| Generative AI | Summarization, content generation, advanced transcription, or multimodal input | Varies by use case | Handles text, images, videos, and audio; choice depends on latency and cost |
Domain 2 of 5
36 practice questions
Practice this DomainLast updated