Three years after her first appearance, Bíró Ada — the AI presenter created for Delta — has evolved from a technology demonstration into a media-psychological phenomenon that provokes emotional reactions and attachment among some viewers. In the latest episode of the podcast “A nagy AI-sztori,” television producer and Ada’s creator Urbán Ernő spoke with communication researcher Aczél Petra, professor at Széchenyi István Egyetem, about how the AI presenter works and why audiences sometimes treat the avatar as a person.
How it works: an avatar, not autonomous intelligence
Urbán Ernő emphasized that Bíró Ada is not an autonomous artificial intelligence but a graphical model: a digital character continuously produced by a 10–15 person team. Editors, graphic artists, technical staff, psychologists and various AI tools jointly create the on-screen persona. Ada does not possess consciousness, does not think independently, and does not react as a human in real time — scripts are written and validated by humans and AI tools are used for animation, voice and visual presentation.
Design choices: why female and why imperfect
The team decided on a female voice and appearance because that combination worked better for the scientific content they present. Early versions looked too sterile and artificial, so the developers intentionally reduced facial symmetry and visual perfection to avoid an overly clinical look. Aczél Petra noted that people still detect a certain sterility in many AI-generated faces; the issue is not merely the absence of errors but a hard-to-describe texture that natural human faces carry.
The name choice was deliberate: “Ada” honors Ada Byron, considered one of the first programmers, while the surname “Bíró” was chosen because it sounds suitably Hungarian and international.
Advantages and limits
Ada can be more effective than a human host in several contexts — she can "travel" to historical settings, be placed in any era, and speak multiple languages. However, in live conversations, programs that require fast reactions, or emotionally charged situations, human presenters remain stronger: AI currently cannot reproduce the spontaneity and live responsiveness that build genuine attachment.
Ethical and societal questions
A central theme of the discussion was the human response: when do we start to form attachments to a digital character, when do we accept it as a person, and when do we lose the ability to distinguish? Urbán Ernő stressed the importance of human oversight: every utterance is backed by validated and professionally checked content. As avatars become more lifelike, the bigger challenge may be how much we want to believe these characters are "alive."
Podcast timestamps and topics
The conversation addressed several concrete topics, with timestamps given in the recording:
- How Ada was created (05:20)
- Where ethical boundaries were drawn for the avatar’s personality (09:07)
- Why even the developers began to feel attachment to Ada (10:53)
- Whether Ada exists as a chatbot (12:30)
- Tasks where an AI-avatar is better than a human (14:35)
- Why people invited the AI presenter on dates (17:42)
- Who is responsible if the AI-avatar makes a mistake (22:16)
- Whether flesh-and-blood presenters are threatened (27:53)
- Legal and ethical limits of AI-avatars (33:15)
Photo: Bíró Ada / Facebook
The episode underlines that Ada’s existence is not only a technological matter: the real questions are how people react to such characters, and how society will shape rules, responsibilities and expectations for AI-driven media figures.
