When TikTok introduced the ability to convert text into AI voice, it was adopted by millions of users around the world. Creators who don't have time, lack recording skills or spoken word command, or who simply prefer to remain anonymous, suddenly have the ability to easily add clear, confident audio to their videos.
But research from the UBC Sauder School of Business shows that AI voice is more than just a useful new tool: it's fundamentally changing the types of videos users create.
Researchers used machine learning algorithms to analyze a dataset of 554,252 videos, including the length of the video, how novel the video was, whether the creator revealed their identity, and whether the overall sentiment of the video was negative. I investigated whether it was positive or not. They also investigated whether users of AI voice became more or less productive.
They found that creators who used AI voice produced 24% more videos per week, and their videos were watched for 63% more time per week. Creators with less experience and popularity were significantly more productive.
“Creating a video is a very complex process and involves many elements, including the topic, storyline, video, and audio,” said study co-author Jean Mu Li, UBC Sauder Associate Professor. . “If you can delegate one task to an AI, you can devote your cognitive resources, time, and energy to other factors. Therefore, your productivity will increase.”
The study also shows that AI voice produces more novel videos, that is, videos that are different in terms of text, visual, and audio content. This effect is more pronounced for experienced creators, perhaps because they have the skills to exploit the full potential of new tools.
AI voice also provides anonymity, which can lead to more negative opinions.
“After people adopt this tool, the sentiments they use in their videos tend to be more negative, and creators tend to appear less frequently in their videos, so they hide their identities more.” ” said Mi Zhou, assistant professor at UBC Sauder. , co-author of the study with Lee and UBC Sauder doctoral student Shaok Zhang.
“We think this is due to a disinhibition effect; your identity is not tied to the content, so you can be more open,” Lee said. The researchers also found that the effects lasted. Even when users returned to using their voices, sentiment remained even more negative.
Interestingly, Zhou said research into AI voices is becoming increasingly difficult, as computer-generated voices have evolved to the point where they are almost indistinguishable from human voices. Next, researchers plan to investigate the impact of her AI voice on consumers to determine whether the use of AI-generated voice changes consumer responses to video content.
“Simple tools can have a big impact. Sometimes the effect is intended, such as increased productivity, but sometimes it's unpredictable,” said Lee, who is currently working with co-authors on the study. It is working on an AI-based design project with South Korean automaker Kia Motors.
Chou agreed, adding that intervention by social media platforms, policymakers and regulators is essential to ensure transparency. “It's important for platforms and policymakers to come up with effective ways to help people distinguish between what content is AI and what content is human. People have a right to know,” she said. Told.
Recently, TikTok has taken steps to label AI-generated content, and just last week, TikTok joined the Coalition for Content Provenance and Authenticity (C2PA) to become the first video-sharing platform to implement content authentication technology. ) has introduced a partnership with
“People are worried about 'deepfakes' that can cause trust issues on online platforms,” Lee said. “I think these are important steps that TikTok is taking to build trust and transparency in the creator economy. However, with the rapid development of AI technology, AI-generated content can be automatically identified. There will be technical challenges in doing so.”
The paper “How will AI-generated audio impact online video creation? Evidence from TikTok” was presented at the 2023 Conference on Information Systems and Technology.

Featured researchers
Shumi
Accounting Information Systems Department Assistant Professor

Featured researchers
Jean Mu Li
Assistant Professor, Department of Accounting Information Systems