Ethical challenges with Generative AI

The Evolving Role of Virtual Journalists and Social Media Influencers

By Gabrielle CHOU - January 20th, 2024

The Chinese Digital Human market is estimated to be worth over 36 billion yuan (US$ 5 billion) in 2023.

Recently, during an interview with Laure Closier Closier from BFM Business TV “La Matinale,” I was questioned about an AI Avatar presenting the Evening News on an American network. Initially, my view was skeptical; the Avatar, despite its realistic 15-second clip, seemed far from delivering substantial reflection or articulating ideas on par with real news anchors. But, upon returning to Asia from the US, my perspective began to shift. Observing the progress of AI Influencers and Virtual Key Opinion Leaders (KOLs) in “Livestream” shows, it’s clear we are edging closer to a new realm of possibilities in media, where the line between artificial and human-driven content is rapidly blurring.

The initial encounter with an AI avatar is deceptively realistic. Within the first seconds, it’s hard to believe that it’s not a human delivering the news and even if a closer inspection doesn’t, a critical question arises: How long will it be before these avatars become indistinguishable from human anchors for extended periods? Here are some specific examples and points of contention:

Social Media Platforms Strategic Focus: In October 2023, Meta invested in AI characters with stars like Snoop Dogg, Tom Brady, MrBeast, and Charli D’Amelio to create their fictional AI characters. emphasizes the shift in content creation, prioritizing digital replication and potentially overshadowing human uniqueness.
Ethical Dilemmas in AI and Celebrity: The use of celebrity likenesses for AI avatars blurs ethical lines, particularly concerning content originality and ownership.
Realism vs. Authenticity in AI Avatars: With AI avatars becoming more lifelike, a debate intensifies over the value of human-led jcontent and its intrinsic qualities like empathy and spontaneity.

However, the landscape is even more dynamic when we look at live streaming and the burgeoning success of AI influencers in Asia:

The Rise of Virtual KOLs in China: In China, virtual Key Opinion Leaders (KOLs) have revolutionized the livestream and influencer landscape. For instance, #Ayayi, a hyper-realistic virtual KOL, has mesmerized audiences with her lifelike appearance and has successfully collaborated with luxury brands. Meanwhile, #Angie’s imperfections and relatable lifestyle have garnered her a substantial following. Notable virtual personalities contributing to this growth include Luo Tianyi, a character-based digital avatar known for her anime-inspired singing persona, Tianyu, who represents the historical fantasy genre, and on the functionality side, Gong Jiuyu serves as a brand ambassador, and Xiao C operates as a digital news anchor.

The numbers say it all: The digital human industry in China is experiencing a significant surge, underscored by the following key figures:

The Chinese Digital Human market is estimated to be worth over 36 billion yuan in 2023.

Investments in the sector exceeded 2.5 billion yuan in 2022.

The industry comprises roughly 317,000 companies as of 2023.

Projections indicate that the market could triple in size within three years.

Chinese consumer engagement is robust, with a 2023 survey finding that nearly 90% of participants had heard of or interacted with virtual personas, and a quarter of respondents from a 2022 survey were open to increasing their spending on digital human interactions in the future. These statistics underscore the rapid digital transformation taking place in China and the significant role that virtual humans play across various sectors.

The unique charm and connection a human journalist or influencer brings are irreplaceable, but AI is rapidly changing the rules of engagement.

The implications are vast for journalism, marketing, and digital content creation:

Copyright and Image Rights: The use of digital likenesses and the creation of entirely new virtual personas present complex legal challenges.
Authenticity of Content: In an era where trust in media is fragile, the introduction of AI avatars could either complicate or enhance the audience’s perception of news and brand authenticity.

This disruptive phase in our industry and beyond raises many questions and possibilities. How do we embrace these changes while maintaining integrity and authenticity? What content will matter to the audiences?

Journalism

Digital Transformation

Live Streaming

Influencer Marketing

Yuval Noah Harari (Sapiens) versus Yann Le Cun (Meta) on artificial intelligenceEXCLUSIVE. The annihilation of democracy or new age of Enlightenment? Best-seller Sapiens author Yuval Noah Harari and Meta head of research Yann LeCun debate.

Transcript: Senate Judiciary Subcommittee Hearing on Oversight of AIThe Subcommittee on Privacy, Technology and the Law hearing was titled "Oversight of A.I.: Rules for Artificial Intelligence."

The Ezra Klein Show - A Skeptical Take on the A.I. Revolution on Stitcher

Machines that think like humans: Everything to know about AGI and AI Debate 3For 3.5 hours, scholars led by Gary Marcus and Vincent Boucher discussed what tools might bring about the fabled 'Artificial General Intelligence,' and what the practical and moral implications might be.

How to create, release, and share generative AI responsiblyCompanies including OpenAI and TikTok have signed up to a new set of guidelines designed to help them be more transparent around generative AI.

How should AI systems behave, and who should decide?We’re clarifying how ChatGPT's behavior is shaped and our plans for improving that behavior, allowing more user customization, and getting more public input into our decision-making in these areas. OpenAI’s mission is to ensure that artificial general intelligence (AGI)[1] benefits all of humanity. We therefore think a

The Alignment Problem from a Deep Learning Perspective (research Paper)

Within the coming decades, artificial general intelligence (AGI) may surpass human capabilities at a wide range of important tasks. We outline a case for expecting that, without substantial effort to prevent it, AGIs could learn to pursue goals which are very undesirable (in other words, misaligned) from a human perspective. We argue that AGIs trained in similar ways as today’s most capable models could learn to act deceptively to receive higher reward; learn internally-represented goals which generalize beyond their training distributions; and pursue those goals using power-seeking strategies. We outline how the deployment of misaligned AGIs might irreversibly undermine human control over the world, and briefly review research directions aimed at preventing these problems.

Research available here