Clinical training & systems

Biased Oracle

Doctors need to understand when to trust an AI recommendation. Play through clinical cases, look under the hood and see what changes when a patient is poorly represented in the training data.

Explore the game ↗Watch the team ↓

THE IDEA

Doctors want to understand how a drug or device works before using it. An AI recommendation needs the same scrutiny: where did it come from, and when might it fail?

Biased Oracle uses clinical scenarios to make those questions playable. Compare the patient, the recommendation and the data behind it. The aim is an intuitive mental model of AI, including the effect of gaps in its training data.

Adapted from original team trailer and submitted deck, with Sam’s summit context.

Gameplay

1 video

Biased Oracle · original team trailer video preview — Biased Oracle · original team trailer58 sec · Open video ↗

The team presentation

1 video

Team presentation video preview — Team presentation198 sec · Open video ↗

Supplemental capture history

The original team trailer is shown in Gameplay above. This earlier agent-created capture is retained as a supplement.

Video history

1 video

Earlier agent-created capture · supplemental video preview — Earlier agent-created capture · supplemental70 sec · Open video ↗

← Discover more projects