- Users can earn small credits while guiding AI to complete everyday tasks.
- The initiative aims to address the challenge of translating complex human intent into reliable AI actions.
Alibaba’s agentic AI app Qwen kicked off a weeklong AI trial on March 30, inviting users to test voice-activated recharge and ride-hailing services, marking a key step in refining its newly launched “AI tasking” capabilities through real-world feedback.
Qwen said it will invite up to 1 million users per day from March 30 to April 6 to participate and provide feedback, offering 12 yuan ($1.75) in ride-hailing credit and 5 yuan in mobile recharge credit on a first-come, first-served basis.
Users simply need to say, for instance, “Top up my phone with 200 yuan,” to complete the transaction, Chinese media reported.
Media reports also indicate that participants who provide valid evaluations may receive an additional round of credits.
The Yangtzeer‘s reporter tested the service, successfully topping up 50 yuan via voice command and receiving 12 yuan in ride-hailing credit.
Although the experience was seamless, the reporter was told that credits will be confirmed and issued within three days after the event and must be used on the day they are received.
Moving forward, Qwen plans to expand AI tasking to more proactive functions, such as remembering family phone numbers, automatically topping up balances, or scheduling monthly recharge.
Since the start of 2026, Qwen has rolled out multiple AI tasking capabilities including AI-powered ride-hailing and ticket booking.
Yet, translating highly personalized, complex real-world needs into accurate AI execution remains a major hurdle. The company believes incorporating direct user feedback is essential to reducing decision errors when AI interacts with the physical world.
