TikTok is considered the meeting place of Generation Z, and since ChatGPT can help with homework, AI on the platform has boomed. Enthusiasm outweighs criticism, according to a study.
ChatGPT has generated a considerable amount of content on TikTok in a very short time – and that’s not counting the AI-generated scripts. Researchers at Ludwig Maximilian University in Munich have now studied how the creators of the most popular videos on ChatGPT portray the chatbot.
Given TikTok’s predominantly young audience – around 40 percent are under the age of 24, according to most statistics – it is revealing to see which aspects of the AI are emphasized and which are not. It may be possible to see early trends in how younger generations, growing up with these entirely new tools, will interact with them.
#ChatGPT: Viewed more than 1.3 billion times
The statistics team scanned and categorized the top 100 most popular #ChatGPT videos in English. In total, videos using the hashtag have been viewed more than 1.3 billion times. The top 100 have a combined total of 250 million views and 22 million likes.
The largest share, 53 percent, is in the “promotional” category, which ChatGPT portrays as mostly positive. Most of these videos give a brief overview of basic AI capabilities, such as answering questions or writing text. Only a few of the videos analyzed go into depth.
Compared to the 53 positive videos, the researchers found only 12 critical videos in the top 100, half of which were about AI detectors like GPTZero. TikTok users might be mostly interested in videos about detectors because they fear getting lower grades on their AI-generated homework if they get caught.
Critical videos have higher interaction rates but lower overall visibility
In addition to the videos themselves, the researchers also tracked the corresponding views, likes, and shares. The average critical TikTok video was viewed more often and received more likes than the average promo video, but was shared less often. Overall, however, the promo videos were viewed, liked, and shared by “much more people” than the critical ChatGPT videos.
The lack of discussion around limitations related to the use of ChatGPT in the analyzed TikTok videos is a concern that requires further exploration. From what we have seen, the risks of producing eloquent but incorrect responses and similar limitations of ChatGPT like inherent biases of the produced output are currently very underrepresented on the most popular social media platform among younger people.
Haensch et al.
The researchers want to encourage teachers to take a critical look at AI tools like ChatGPT in the classroom, rather than simply banning them. After all, students would find out about it anyway – on TikTok, for example.