RAIUC03-BP01 Identify the expected input and outputs for the AI system

Define your AI system's inputs at a high level by describing the types of information that will flow into your system, such as text in multiple languages, images, audio recordings, or structured data like user preferences or transaction histories. Focus on the content and meaning rather than technical formats. For example, you could define inputs as customer support conversations rather than MP3 audio files.

Specify what your AI system should produce as outputs, describing the type of information it will generate. This might include text responses, classification labels, recommendations, generated content, or structured data that downstream systems can use to take actions.

Analyze whether your inputs contain enough information to reliably produce your desired outputs by thinking through the logical connection between what you're giving the system and what you expect it to produce. If you want your system to diagnose medical conditions but only provide it with basic symptoms, consider whether that's sufficient or if you need additional input like medical history or test results.

Compare your AI system's inputs and outputs to how the problem is solved today without AI, identifying what's different and what stays the same. For example, if human customer service agents currently handle inquiries using phone calls and internal knowledge bases, but your AI will work with chat messages and documentation, note these differences and consider their implications.