NASA’s new dark energy space telescope can also detect killer asteroids
technologyreview.com·2h ago
Data Science
Simon Willison's Blog·Jul 3·5 min read
Fable's judgement
TL;DR
Software engineers often struggle with how to effectively utilize AI models like Fable for testing and coding tasks. Instead of imposing strict guidelines, it's more effective to allow Fable to use its own judgement in deciding when to run tests and which models to apply.
✦ Why It Matters
Engineers can improve project efficiency by allowing AI models to make judgement calls on task management.
Key Takeaways
- Allow Fable to use its judgment for task delegation.
- Delegate simpler coding tasks to lower-power models.
- Keep complex tasks with the main AI model.
- Monitor resource usage to optimize efficiency.
How It Works
Fable can assess the complexity of tasks and decide whether to use a lower-power model for simpler coding tasks. This delegation allows for efficient resource management while maintaining high-quality outputs for more complex tasks.