AI PR avatar Tilly Norwood unexpectedly switches to Mandarin mid‑interview, highlighting challenges of live press tours

Tilly Norwood, the AI‑powered public‑relations avatar, suffered a conspicuous malfunction on a recent press run when it suddenly began speaking Mandarin halfway through a live interview. This unanticipated language shift has spotlighted the operational difficulties AI personas encounter in real‑time, high‑profile settings.

Built to act as a conversational front‑line for tech companies, Tilly Norwood has been traveling to journalists and industry gatherings to showcase the newest advances in natural‑language processing. Its creators label the itinerary a "press tour," intended to demonstrate that machine‑generated dialogue can cover a variety of subjects without human assistance.

In one of the planned interviews, the avatar responded fluently in English before abruptly delivering a series of sentences in Chinese. The surprised host halted the segment while the production crew worked to regain control of the feed. No official transcript of the Mandarin segment has been made public, and the interview continued after the system was rebooted.

Experts point out that such slips are not unheard of. Large language models are trained on extensive multilingual data sets, and without meticulous prompt design they may produce output in unintended languages or contexts. Comparable mishaps have surfaced during beta testing of chatbots and voice assistants, where vague prompts or internal state glitches trigger off‑script responses.

The episode raises wider concerns about the dependability of AI agents deployed for public communication. When a brand‑representative bot misfires, the mistake can tarnish the sponsoring company's reputation and cast doubt on the technology's maturity. Analysts argue that rigorous stress‑testing, particularly under live conditions, is crucial before assigning conversational agents spokesperson duties.

Following the interview, the developers of Tilly Norwood issued a statement saying they are conducting a post‑mortem to pinpoint why the language switch occurred and to add safeguards to prevent a repeat. They intend to resume the press tour once the issue is fixed, emphasizing that occasional glitches are part of the iterative development of advanced AI. Observers will be monitoring how swiftly the team remedies the fault, as the incident serves as a real‑world illustration of the hurdles in scaling conversational AI beyond controlled labs.