October 2023
Tinius Digest
Monthly reports on changes, trends and developments in the media industry.
About Tinius Digest
Tinius Digest report on changes, trends and developments within the media business at large. These are our key findings from last month.
Share the report with colleagues and friends, and use the content in presentations or meetings.
The dearth of news on TikTok
Researchers at Northwestern University have analysed the spread—or lack thereof—of news content on TikTok.
Key findings:
1
Limited news coverage
There is a marked lack of interest in news content on TikTok. Most news organisations on the platform only manage to gather a modest audience, and the For You Page algorithm barely surfaces news content, even when influenced by active engagement signals.
2
Minimal recommendations
An analysis of 6,568 videos from the For You Page revealed that only six videos, comprising just two unique pieces of content, were somewhat news-related. This finding indicates a significant underrepresentation of current events and news-related content in TikTok's algorithmic recommendations.
3
Entertainment-dominated
The top trending hashtags on TikTok primarily focus on entertainment, with little to no representation of news events. Categories like film, television, and music were predominant, while news-related content was notably absent from the list of most popular hashtags.
4
Low engagement for news media
Among 120 news media profiles analysed, the median account had 256,500 followers, with 78 percent of accounts having fewer than one million followers. This indicates that even TikTok's most prominent news accounts only reach a small fraction of the platform's user base in the United States.
Sweden: A new (digital) normal
The Swedish Press and Broadcasting Authority has published its annual report on media consumption and trends in Sweden.
Key findings:
1
Elevated post-pandemic news interest
The pandemic saw a spike in news consumption and newspaper reading, with levels remaining high post-pandemic. Traditional news media continues to be the most trusted source.
2
Generational digital divide
While older demographics increasingly use digital media, they still predominantly favour traditional outlets such as radio and TV. In contrast, those in the 15-24 age bracket spend an average of 130 minutes daily on social media, compared to just 27 minutes for the 65-85 age group.
3
Social media for news
Approximately 46 percent of Swedes aged 9-85 years engage with social network services for news, which rises to 76 percent amongst 15-24-year-olds.
4
Trend towards streaming
There is a noticeable shift from traditional TV to Video On Demand (VOD) services. 76 percent of Swedish households have access to video streaming services, with 68 percent subscribing to music or podcast streaming services.
ChatGPT excels in emotion recognition over humans
Researchers from Israel and the UK have evaluated ChatGPT's ability to predict human emotional responses in 20 hypothetical situations, comparing its performance against that of the broader population.
Key findings:
1
Achieving maximum scores
In the initial study, ChatGPT significantly surpassed the general population regarding emotional awareness. A follow-up study a month later showed a remarkable improvement in ChatGPT's performance, closely approaching the highest score attainable in the test. The accuracy level was impressively high, scoring 9.7 out of 10.
2
Progressive improvement
The research demonstrated that ChatGPT can generate appropriate responses in the context of emotional awareness, which is enhanced progressively over time. The researchers emphasize that while ChatGPT can identify and express human emotions, it does not equate to emotional intelligence or empathy.
3
Application in mental health
The study proposes that ChatGPT could be instrumental in cognitive training for individuals facing challenges in emotional awareness. Furthermore, the chatbot might be beneficial in supporting psychiatric diagnostic processes and assessments and enhancing emotional language proficiency.
Widespread disinformation on TikTok in Norway
TikTok has published its semi-annual report for platform moderation in the EU and EEA, providing specific figures for TikTok in Norway for the first six months of 2023.
Key findings:
1
Limited fact-checked videos
Of the videos in Norway, only 341 were fact-checked, with 138 subsequently removed. Notably, 22.4 percent of Norwegian users refrained from sharing unverified content after receiving a content warning from TikTok.
2
Dangerous disinformation
A total of 2,055 videos were identified as dangerous disinformation with the potential to incite violence against individuals or groups. These videos garnered 7,629,680 views. The figures represent videos published in Norway with global engagement.
3
Holocaust and pandemic misinformation
Videos disseminating false information about the Holocaust attracted 6,201,917 views, with 13,128 users engaging with them. Similarly, videos spreading misinformation about the pandemic were viewed 5,049,439 times, with 5,113 user interactions.
4
Political advertising violations
Five political advertisements that breached TikTok's guidelines were removed after accumulating 313,944 views.
5
Fake accounts and engagement
In the period from March to June, TikTok removed 27,052 fake accounts, amassing a total of 1,284,990 followers. Additionally, 797,166 fake likes and 580,625 fake followers were eliminated. Furthermore, 166 profiles were suspended for impersonating others, including individuals, businesses, or brands.
Not necessarily more efficient with AI
Boston Consulting Group has published the result of an internal study (n=758) of the effects of AI on knowledge worker productivity and quality.
The consultants were divided into three groups: one with no access to AI tools, one with access to AI tools, and one with access to AI tools plus guidance on effective prompt usage.
Key findings:
1
Less AI knowledge—less efficient
Efficiency gains from AI tools are not automatic. Consultants lacking training and prior knowledge of AI tools showed lower efficiency than their trained counterparts.
2
Challenges of AI tool adoption
The main challenges leading to decreased efficiency included a lack of understanding of AI tool limitations, blind reliance on AI data, a significant learning curve, and overdependence on AI tools, impairing critical thinking and problem-solving skills.
3
Positive impact with proper use
When used effectively, AI tools led to notable improvements: a 12 percent increase in task completion rate, 25 percent faster task completion, and 40 percent higher quality of work than those not using AI tools.
AI used for disinformation in 47 countries
Freedom House has published its annual 'Freedom on the Net' report. No Nordic countries are included in the report.
Key findings:
1
Continuous decline in internet freedom
Global internet freedom has decreased for the 13th consecutive year, with severe repression in Iran, Myanmar's near-bottom ranking, and deteriorating conditions in the Philippines and Costa Rica.
2
Rising attacks on free expression
In a record 55 out of 70 countries, individuals faced legal consequences for online speech, with physical assaults and fatalities linked to online commentary in 41 countries. Notably, harsh punishments were observed in Myanmar and Iran.
3
AI-driven disinformation campaigns
AI technology is increasingly used for online disinformation, with at least 47 governments manipulating online discourse. AI tools capable of generating text, audio, and imagery are being utilized in 16 countries to undermine truth and influence public opinion.
4
AI-enhanced censorship
Authoritarian governments are leveraging AI to intensify online censorship, with 22 countries employing AI for content removal. Despite AI's rise, traditional censorship methods persist, as evidenced by 41 governments blocking websites adhering to international free expression standards.
5
Call for responsible AI governance
To safeguard internet freedom, democratic supporters must establish strong, human rights-based standards for AI development and use, addressing the amplified risks of digital repression, surveillance, and disinformation through AI technologies.