Meta's Muse AI Agents Use Human Staff for Some Functions
Meta's Muse AI Agents Incorporate Human Assistance
Meta is advancing its AI-powered Muse agents, designed to handle tasks like business inquiries and bookings on behalf of users. However, reports indicate that the company is supplementing these AI interactions with human staff—a strategy reminiscent of its discontinued 2015 M AI project.
Hybrid Approach in Early Rollout
According to a report by 404 Media, Meta’s Chief AI Officer Alexandr Wang confirmed last week that the company expanded Muse’s outbound calling capabilities. While the AI is intended to autonomously manage tasks such as pricing queries and reservations, leaked internal insights reveal that some calls are being routed to trained human agents.
This hybrid model aims to ensure smoother functionality during Muse’s early testing phase. Human intervention helps address potential AI shortcomings while simultaneously refining the system.
Echoes of Meta’s Past AI Experiment
Meta previously attempted a similar approach with M, an AI assistant launched in 2015 within Messenger. Like Muse, M combined AI automation with human oversight to fulfill user requests, including restaurant bookings. However, scalability issues and high operational costs led to its shutdown in 2018.
Casey Newton of The Verge noted limited user engagement with M, a sentiment echoed by Meta CEO Mark Zuckerberg. Despite past challenges, Meta is revisiting the hybrid model, betting on improved AI advancements to make Muse more viable.
Privacy Concerns and Future Outlook
Reuters reported that Meta tested a "human concierge" system to enhance AI responses but paused the initiative over data privacy concerns involving third-party contractors. The company remains committed to AI-driven solutions, though the balance between automation and human oversight remains a critical challenge.
Like.tg will continue monitoring developments as Meta refines its AI strategy.
Meta’s 2015 M AI project also relied on human trainers for complex tasks.
