AI can be powerful, but it’s never flawless. Imagine an AI system supporting hospital staff in making urgent decisions. A hallucination, contradiction or wrong answer could directly affect safety, even lives.
Provide a solution on how you would approach and solve this. Describe a solution in which hospital staff would be supported by AI decision(s). Showcase how you’d evaluate AI decisions, put the human in the loop, and counter adversarial input on your model.
Feel free to include potential Python frameworks to work with or describe an (agent) architecture, but be sure to provide a plan of action explaining how you’d deliver results that would pass evaluation and remain stable.
