Brind.

Meta and Google Test Human Oversight for Advanced AI Agents

1 report, 1 independent Updated Sep 23
AI-generated briefing. Brind wrote this from the reports listed below. It can be wrong. Each section says how much you can rely on it, and the sources are linked so you can check.

What happened

Some supportReported by 1 outlet

Artificial intelligence agents are being developed to handle tasks such as booking appointments, online shopping, and making phone calls without constant human guidance. Meta is currently testing a 'human concierge' for its personal AI agent, Muse, where human contractors handle calls to US businesses that Muse cannot complete on its own. The challenges of agent autonomy are also evident in other systems, including Google's Duplex and Perplexity, which have utilized human operators during their testing phases.

From business-standard.com

Why it matters

Some supportBrind's analysis of the reports

The testing highlights the gap between performing tasks in a controlled environment and executing them successfully in the unpredictable real world. This reliance on human intervention remains crucial when AI agents interact with external factors like human businesses or websites that do not behave as expected.

From business-standard.com

Who's involved

  • MetaTesting a personal AI agent called Muse, which utilizes a human concierge service for certain calls.
  • GoogleTheir AI agent Duplex used human operators when facing specific task limitations.

Keep exploring

The entities involved

Related events

Coverage

Newest first; wire copies grouped