Framework Maps Three AI Roles for Human-Machine Collaboration in Autonomous Vehicles

A new framework outlines how artificial intelligence and humans can work together more effectively in transportation systems, moving beyond existing automation standards that focus solely on which party controls the vehicle.

The Limitation of Current Standards#

Today's automation frameworks, including the SAE Levels of Automation, primarily define roles based on who sits in control. Yet they leave a critical gap unanswered: how should humans and AI actually collaborate moment-to-moment, especially when driving conditions shift rapidly and responsibilities need to change hands?

Three Dynamic AI Roles#

The researchers introduce a model with three distinct AI personas that adapt based on real-time conditions:

• Advisor: Provides information and recommendations without taking direct action
• Co-Pilot: Offers partial control sharing and intervenes when needed to assist the driver
• Guardian: Steps in with emergency override capabilities to prevent harm

These roles aren't static. Instead, the AI switches between them based on factors like the driver's mental state, environmental hazards, and immediate vehicle conditions.

Practical Application and Trust#

The framework demonstrates its potential through e-scooter use cases, showing how these distinct AI behaviors influence user confidence and acceptance. By tailoring intervention levels to user needs, the model aims to build trust while maintaining safety in shared-control scenarios.

Step Toward Universal Standards#

The work represents an early effort to create a role-based collaborative standard for AI across transportation systems—from fully automated vehicles to micromobility devices and remote vehicle operation. The researchers emphasize designing AI systems that embed smoothly into future mobility while respecting human agency and interaction patterns.