DoorDash's Tasks App: Gig Workers Paid Pennies to Train AI — The Bleak Future of AI Labor

DoorDash's Tasks App: Gig Workers Paid Pennies to Train AI — The Bleak Future of AI Labor

DoorDash's new Tasks app allows gig workers to earn money by recording videos of themselves performing everyday activities — doing laundry, cooking, cleaning, walking — to train AI models. The experience reveals a disturbing vision of the future where human labor is reduced to training data for machines that will eventually replace it.

How It Works

The Tasks app assigns gig workers specific activities to record:

• Video recording: Workers use their phone cameras to film themselves performing tasks
• scripted activities: Doing laundry, scrambling eggs, walking in parks, organizing items
• Micro-payments: Workers earn small amounts for each completed recording
• AI training: The videos are used to train computer vision and robotics models

The Bleak Reality

A WIRED reporter's first-hand experience revealed:

• Minimal pay: Recording mundane activities for very low compensation
• Surveillance feeling: The constant camera recording creates an uncomfortable surveillance dynamic
• Redundancy: Workers are essentially demonstrating tasks that AI will learn to perform autonomously
• No career path: This type of gig work offers no skills development or advancement

The Business Logic

DoorDash's strategy makes economic sense:

• Robotic delivery: DoorDash is investing in autonomous delivery robots
• Training data: Human demonstrations are the most effective training data for robotics
• Cheap labor: Gig workers are paid far less than professional data annotators
• Scale: Millions of potential gig workers available through existing DoorDash platform

The Bigger Pattern

This is part of a broader trend:

• Scale AI: Pays workers to create training data for AI models
• Tesla: Uses fleet data from human drivers to train autonomous driving
• Data labeling: Millions of workers globally perform micro-tasks for AI training
• The paradox: Workers who train AI are training their potential replacements

Ethical Questions

1. Informed consent: Do workers understand they're training their replacements?
2. Fair compensation: Is the pay proportionate to the value created for the AI company?
3. Dignity: Does reducing human activity to AI training data dehumanize work?
4. Regulation: Should there be disclosure requirements for AI training labor?

Source: WIRED | Full Report