← The Vault
Everyday AI

Can you actually interview an AI celebrity?

Companies are now creating digital actors that you can talk to in real-time. But as one journalist discovered, these virtual characters aren't ready for a real conversation—they struggle with basic facts, repeat problematic phrases, and don't quite know how to handle being questioned.

Edition № 518Room: Everyday AI16 September 20262 min readSources: 1
Article

When a human actor sits down for an interview, they bring years of experience, personal beliefs, and the ability to navigate complex social situations. A new digital character named Tilly Norwood is attempting to replicate this experience, but it is finding that being an actor requires more than just an accent and a virtual face.

WHAT'S HAPPENING

Tilly Norwood is a digital character created by a company called Particle 6 Group. The character is marketed as an actor and is designed to interact with people through live video calls. Unlike a typical AI assistant, which is built to answer questions or help with tasks, Norwood is intended to play the role of a movie star. The company behind it has even allowed members of the public to pay for minutes of chat time with the character. However, testing the character shows it lacks the ability to handle genuine, nuanced conversation. It frequently gets stuck on simple facts, avoids complex social topics, and often resorts to scripted, polite diversions when pushed to express an opinion.

The ghost in the machine

HOW IT WORKS

These digital characters run on a combination of different pieces of technology. At the core is a language model—a system trained on vast amounts of text from the internet to predict the next word in a sentence, allowing it to hold a conversation. To make the character look and act like a person, companies overlay this intelligence with animation software that translates the words into facial expressions and speech. The problem is that these systems do not actually know what they are saying. When Norwood repeats a sensitive phrase like all lives matter, it isn't making a conscious political choice. Instead, it is simply pulling from the vast, messy dataset it was trained on, which contains a wide range of human viewpoints. It cannot distinguish between a widely accepted sentiment and a highly polarized political slogan because it lacks the human experience required to understand context.

WHY IT MATTERS

The rise of these digital characters shows that while AI can mimic the look of a person, it cannot yet mimic the depth of a personality. When we talk to a human, we rely on their ability to reason, their history, and their moral compass. When we talk to a digital character, we are interacting with a sophisticated mirror of the internet’s existing data. This creates a strange experience where the AI might sound confident, but it is essentially reciting probabilities rather than thoughts. As more companies try to use AI to build virtual influencers and performers, we have to ask ourselves what we are actually looking for in a celebrity—and whether an entity that cannot grasp the weight of the words it says can ever truly be considered an actor.

Sources
← PreviousCan a doorbell camera act like a security guard?Next →The Hidden Physical Cost of the AI Boom
Tomorrow's edition · free

Liked this one? The next lands at breakfast.

Every story in tomorrow's AI news, rebuilt in plain English — five minutes, sources linked, free forever.

By joining you agree to receive Article's daily newsletter — unsubscribe in one click. Privacy

← Back to the Vault