Keep the choice yours when you delegate to AI

A useful first step with AI isn’t handing over more—it’s deciding what you actually want to hand over.

Honeywell’s announcement about Forge Cognition describes manual, semi-autonomous and autonomous operating modes. That’s a company description, not independent validation of the reported results—but the choice of oversight is a useful prompt for our own decisions.

Try this with one low-stakes task for a week:

• What may the tool suggest or prepare?
• What do you want to review and approve?
• What decision do you want to keep yours?

Then notice whether the arrangement gives you room for what matters, or just adds another thing to monitor. NIST’s AI Risk Management Framework also emphasizes defining human roles and oversight in context; it doesn’t prescribe one level for every use.

Sources: Honeywell’s announcement · NIST AI Risk Management Framework