At Ars Technica, we’ve been keenly observing the intriguing potential and substantial risks, including occasional oddities, of AI-generated videos that depict increasingly lifelike human avatars. Our journey with this technology dates back to when we created a deepfake of Mark Zuckerberg in 2019, noting the advancements in AI since then.
Despite our extensive background, the concept envisioned by startup Channel 1 of AI-generated avatars delivering news was quite startling. Their recent proof-of-concept newscast demonstrated the rapid evolution of AI-generated human representations and the profound implications for industries beyond just broadcast journalism.
To clarify, Channel 1 does not aim to deceive viewers with deepfakes of existing anchors. In their sample broadcast, they explicitly identify their presenters as “AI-generated reporters” and disclose that no real person recorded the presented speech. Despite such disclaimers, it’s challenging to remember that these are not real people but avatars modeled after individuals who were compensated for their likenesses.
The AI avatars from Channel 1, from their lip syncing to their gestures, present an uncannily life-like depiction. While careful scrutiny reveals subtle flaws that give away their artificial nature, these are minor enough that they might go unnoticed, especially on smaller screens or during a cursory view.
Human-like AI avatars are seemingly overcoming the ‘uncanny valley’ in specific applications like news delivery, where the anchor is stationary. Channel 1’s technology has improved to the point that it’s becoming comfortable for viewers to watch these AI-generated newscasters.
Channel 1 also showcases the potential for on-location news application, offering an example where AI technology translates a French-speaking storm victim’s account into English, retaining the original voice and syncing the lips accurately. This application transcends typical TV news translation methods and could be easily mistaken for an authentic English-speaking source if not for the disclaimer highlighting the AI translation
Read More
Seguridad en la Nube y AWS La seguridad en la nube es un elemento esencial…
Ciberseguridad y TI Hacking Ético Ciberseguridad La ciberseguridad es fundamental en el mundo actual donde…
Introducción a la ciberseguridad y TI Hacking etico ciberseguridad En el mundo digital de hoy,…
Introducción a la ciberseguridad y TI servicio de IAM Entornos Multicloud En el mundo actual,…
# Beneficios de seguridad de la computación en la nube ## Introducción La computación en…
Brechas de seguridad en la computación en la nube La computación en la nube ha…
This website uses cookies.