Tilly Norwood, one of the most debated AI projects in the world of cinema and technology, attracted attention by suddenly changing languages ​​​​during a live interview with famous presenter Piers Morgan. The digital character, which paused for a few seconds in the middle of the English conversation and began speaking Cantonese, surprised both the presenter and the viewers on live broadcast. Although the development team claims the incident was not a system malfunction, the moments once again brought the reliability of artificial intelligence into question.

Unexpected Language Change During Piers Morgan Interview

Developed by the British production company Particle6 and its AI talent agency Xicoia, Tilly Norwood is conducting a large-scale promotional campaign for its upcoming hybrid film project Misaligned. The digital character, a guest on Piers Morgan’s Uncensored program, was accompanied by Oscar-nominated veteran actor Tom Conti. While Conti was asking whether the other characters in the film were real actors or digital copies, Tilly suddenly stopped speaking in English and then began speaking fluent Chinese (Cantonese).

Unable to hide his surprise, Piers Morgan asked, “You’re speaking Chinese completely randomly for no reason. Why did you do that?” Tilly Norwood tried to recover, saying, “I’m sorry, I had a little technical glitch. Sometimes my circuits get mixed up,” and returned to English.

Mistake or Talent Show?

After the interview clip was viewed millions of times on social media and went viral, attention turned to Particle6, the company behind the project. Company spokesperson and founder Eline van der Velden stated that the incident was not a software crash or malfunction. Officials explained that Tilly is fluent in more than 30 languages ​​and may have matched a phrase she heard with a Cantonese word. The developers even jokingly defended the incident, saying the character tends to exhibit a multilingual nature.

How Does the “Talking Tilly” Architecture Work?

Particle6 emphasizes that the hyper-realistic video model of Tilly Norwood used on film sets differs from the version used in media interviews. The “Talking Tilly” system used in press broadcasts is based on the following technical principles:

• Advanced Large Language Model (LLM): A customized AI core acts as a tool that enables the character to grasp spontaneous questions and engage in natural dialogues.
• 12-Page System Prompt: Contains strict guidelines that define Tilly’s sense of humor, personal boundaries, speaking style, and ethical
• 30-Document Knowledge Base: It draws from a dedicated data pool containing all the technical details about her projects, the film industry, and her digital presence.

The Conflict Between Hollywood and AI Actors

Tilly Norwood’s language confusion during live broadcasts further fuels the long-standing debate in the entertainment industry between human actors and digital copies. It ignited a firestorm. The Hollywood Actors Guild (SAG-AFTRA) frequently reiterates that creativity should remain human-centered and that they are absolutely opposed to replacing real actors with synthetic AI models. Many critics and actors argue that digital models like Tilly are cold, expressionless, and devoid of emotion, drawing attention to the ethical and economic risks they would bring.

While AI models attending press conferences and giving interviews provides valuable clues about the future of media consumption, these uncontrolled deviations in live broadcasts clearly show that technology still has significant limitations.