Gen Z creators dominate content production. Official data from TikTok shows that by 2025, users aged 16 to 24 will account for 76% of creators with related tags, with an average daily output of 3.2 items per person. However, the median account life cycle of these young creators is only 11 months, which is 37% lower than the average of the entire platform. The main reasons are algorithm fatigue and the pressure of ethical complaints. The German Youth Research Institute tracked 500 accounts and found that for those who used the AI smash or pass function more than four times a week, the fan churn rate was as high as 33% per quarter, and content homogenization became the biggest growth bottleneck.
The activity of transgender users has soared. Tubular tracking by YouTube’s data analysis company found that LGBTQ+ creators use this feature 2.1 times more frequently than heterosexual users, among which non-binary users account for 27% (the total user share is only 3.5%). In the documentary about the transgender journey of Brazilian Internet celebrity Carlos, the AI judgment session saw the daily interaction volume soar to eight times the normal level, but the density of negative comments also reached 28% (with a total of 89,000 comments). Dutch media scholars have pointed out that such content has increased the community visibility of Genderqueer by 47%, while triggering a 130% increase in targeted reports.
The technological upgrade of the celebrity fan economy: The AI tool developed by the fan club of Indian Bollywood actress Aliya Bat can score 500,000 celebrity photos in real time, with an average daily request volume of over 1.2 million times. This module has increased the user stickiness of the fan club’s Discord channel to 3.7 hours per person per day, and the membership fee income has grown by 290%. However, legal risks emerged simultaneously – in 2024, the Mumbai High Court ruled that the creator had infringed on their portrait rights and ordered them to pay the star a share of 180,000 US dollars, equivalent to 89% of the annual advertising revenue.
The traffic harvesting engine of marketing agencies, the British MCN company Influenced uses AI templates to mass-produce content, reducing the cost per piece to 4.2 (58 for manual creation). The average monthly play volume of the matrix accounts it operates reaches 700 million times, but the user retention rate is only 0.8% (the industry average is 3.2%). The quality of traffic conversion is even more worried: The product placement case of the beauty brand Fenty shows that the average transaction value of consumers driven by AI evaluation is 23.7, which is 43% lower than the 41.5 converted by refined content, and the return rate is as high as 27%.
A cross-border youth social experiment conducted by Cairo University in Egypt tracked 1,000 users aged 13 to 15 and found that 89% of them used AI tools to evaluate international celebrities. The technical log shows that these teenagers launch an average of 32 cross-border judgments per week, but the cultural misinterpretation rate is 38% (such as misjudging the costumes of Japanese geishas). What is even more alarming is the act of device jailbreaking – in Saudi Arabia, the proportion of users under the age of 18 accessing AI tools through VPNS has reached 22% of the total usage, evading the content review of the country’s Digital ethics committee.
Hololive, a technological ally of the virtual idol industry in Japan, integrated ai smash or pass into its VTuber fan interaction system in 2024, increasing gift revenue by 47%. Its unique three-dimensional evaluation system covers character design (with a weight of 40%), vocal traits (35%), and talent performance (25%), reducing the controversy rate by 58% compared to traditional appearance evaluation. The live-streaming room test data of Chinese virtual singer Luo Tianyi further shows that the AI’s real-time feedback has increased the audience’s reward frequency to 12.3 times per minute and extended the average stay time to 28 minutes.
The current data flow reveals a hidden class gap: In the account matrix operated by the Philippine marketing company 1Social, 85% of the people on camera come from elite families in the capital economic circle, creating a false “landscape of civilian participation”. Ethicists criticize that this mechanism is exacerbating the digital caste system – when the system processes over 14,000 judgments per second, the class bias in its training data will be exponentially magnified. Although technological innovation has achieved a 98.3% interception rate in TikTok’s DeepFake detector, eliminating algorithmic bias still requires interdisciplinary collaboration. Just as Article 29 of the EU’s Artificial Intelligence Act requires content platforms to disclose user profile data sources to deconstruct the power structure hidden behind AI smash or pass.