Generative AI Leader Practice Question: A healthcare company needs to generate synthetic…

Answer analysis

Option-by-option breakdown

For each option: why learners choose it and why it is or isn't the right answer here.

• ✗

  Codey for code generation

  Why it's wrong here

  Codey generates and completes source code, so it produces no pixel data at all and cannot synthesise medical images or enforce patient privacy controls. It is tempting because Codey is a Google Cloud generative AI model, but its modality is code; synthetic medical imaging requires an image-generation model trained on clinical data.

• ✗

  Chirp for speech recognition

  Why it's wrong here

  Chirp performs speech-to-text transcription, consuming audio and returning text, so it cannot produce synthetic medical images or address privacy compliance for imaging data. It is tempting because Chirp is a Google Cloud generative AI service, but its modality is audio; the scenario requires image synthesis, not transcription.

• ✓

  Imagen on Vertex AI

  Why this is correct

  Imagen on Vertex AI generates synthetic images from text prompts, so the company can create research imagery without exposing real patient data. Vertex AI's enterprise controls and data-handling commitments support the privacy compliance constraint, unlike general-purpose image tools lacking healthcare-grade governance.

• ✗

  Gemini 1.5 Pro with multimodal prompting

  Why it's wrong here

  Gemini 1.5 Pro generates and reasons over text, images and audio, but does not train or emit synthetic medical image datasets, nor provide the de-identification pipeline privacy compliance demands. It is tempting because multimodal prompting accepts image inputs, yet that is for analysis and description; synthetic image generation requires a dedicated medical imaging generative model.

Go deeper

Related to this question