Februar 2024
Tinius Digest
Månedlige rapporter om endringer, trender og utviklinger i mediebransjen.
Om Tinius Digest
Tinius Digest gir deg en oversikt over rapporter om og analyser av utvikling i mediebransjen og publiseres en gang i måneden. Her er våre viktigste funn fra denne måneden.
Del gjerne innholdet med kollegaer og bruk det i møter og presentasjoner.
Innhold
- Publishers optimistic despite challenges
- The future of kids' digital lives
- The Nordics are leading in AI adoption
- The key to reader revenue
- Generative AI use divides Norwegian public opinion
- Machine translations dominate low-resource languages on the web
- AI is reshaping journalism and the public arena
- From technological optimism to technological race
Publishers optimistic despite challenges
WAN-IFRA has published The World Press Trends Outlook, which comprehensively overviews the global media industry's financial health, challenges, and opportunities.
Key findings
1
Brighter business prospects
Despite ongoing geopolitical uncertainties, high inflation, and digital disruption, over half of surveyed publishers (55%) are optimistic about their business prospects for 2024. The optimism extends to long-term growth, with 58 per cent of respondents projecting revenue increases in the next three years.
2
Revenue growth and diversification
Media organisations expect a 19 per cent increase in revenue in 2024. Publishers attribute this to diversified income streams, with digital advertising, subscriptions, and alternative revenue sources such as events, grants, and business services all contributing to growth.
3
AI and automation as investment priorities
AI and automation top the list of tech investment priorities, with 87 per cent of respondents planning to invest in these areas. Publishers see AI as key to improving content creation, enhancing workflows, and driving future growth despite concerns about its impact on the industry.
4
Print revenue decline
While digital revenues rise, print advertising and circulation continue to decline rapidly. This constitutes a severe challenge because print remains a significant income source, generating more than half of total revenue in many markets.
The future of kids' digital lives
The 2023 Annual Data Report by Qustodio, a leading online safety provider, analyses the digital habits of children and families worldwide. The report is based on app usage data from over 400,000 families globally, covering critical regions like the US, UK, Spain, France, and Australia.
Key findings
1
Increased screen time and app use
In 2023, children spent an average of four hours daily on their devices, matching last year’s figure. Popular apps included YouTube, TikTok, and Roblox. YouTube remained the top online video platform globally, while TikTok continued to dominate social media use.
2
TikTok and YouTube dominance
Children spent 27 per cent more time on online video platforms compared to 2022, with YouTube and TikTok being the clear leaders. The average daily time spent on TikTok in the UK rose to 127 minutes, highlighting the growing influence of short-form video content.
3
Gaming trends
Gaming remains a crucial aspect of children's digital entertainment, influencing social interaction and learning. Roblox maintained its position as the most-played game among children, with an average daily playtime of over two hours globally.
4
Parental concerns over online risks
Parents ranked exposure to adult content, online predation, and cyberbullying as their top concerns. Parental controls are being widely adopted, with 89 per cent of families using them to block inappropriate content and monitor screen time.
The Nordics are leading in AI adoption
Silo AI has published their third annual *Nordic State of AI* report. The report provides an in-depth analysis of how traditional Nordic companies integrate AI technologies into their operations.
Key findings
1
AI adoption is widespread
While nearly all surveyed companies are experimenting with AI, satisfaction remains divided. About 50 per cent are satisfied with their AI results, while the rest remain neutral. Companies with clear frameworks to assess return on investment (ROI) report higher satisfaction levels.
2
Generative AI and large language models are gaining traction
Although generative AI like ChatGPT dominates headlines, companies also explore various AI technologies, including natural language processing (NLP), optimisation engines, and computer vision. AI is used in products, production, and customer service, especially manufacturing.
3
AI infrastructure and data remain crucial
Data quality and AI infrastructure are vital challenges. Companies must invest in high-quality data and scalable AI infrastructure to ensure successful deployment and long-term benefits. Public cloud solutions are popular, but companies use on-premise resources for AI computation.
4
Talent development outpaces recruitment
Nordic companies invest heavily in training and competence development rather than recruiting AI talent. The demand for data scientists, machine learning engineers, and professionals with domain-specific knowledge continues to grow.
5
Regulation and compliance are significant concerns
As the EU prepares to finalise the AI Act, companies prioritise compliance, with 54 per cent planning to invest in regulatory oversight. Early adopters of AI regulation may gain a competitive advantage by ensuring financial, technical, and procedural compliance.
The key to reader revenue
The World Association of News Publishers (WAN-IFRA) has published 'Best Practice in Reader Revenue'.
Key findings
1
Registrations
Requiring registration provides direct, first-party data that helps media companies fine-tune their content offerings and subscription strategies while reducing churn.
2
Examples of success
The report includes ten examples of initiatives and new products to increase engagement and reach younger audiences. Among the examples is FT Edit, a standalone app from the Financial Times offering access to eight articles per day, designed to reach a younger, more price-conscious audience. Norwegian Amedia is showcased as an example of how digital growth can offset the decline in print.
Generative AI use divides Norwegian public opinion
The Norwegian Data Protection Authority and the Norwegian Board of Technology have commissioned a survey to explore Norwegian attitudes towards generative artificial intelligence.
Read more (in Norwegian).
Key findings
1
Limited AI experience
Over half of respondents (58%) had never tried a generative AI tool. However, usage was higher among younger people, with 79 per cent of those aged 18-29 using such tools primarily for personal rather than professional purposes.
2
Concerns about data privacy
Most (59%) expressed concerns over how AI tools manage personal data. This concern was equally shared by both users and non-users of generative AI tools, highlighting broad apprehension around privacy.
3
Uncertainty about AI’s societal impact
Many respondents were unsure about AI’s overall benefits, with about half agreeing or disagreeing that AI would benefit Norway. Similar uncertainty was noted regarding AI’s potential in education and the workplace.
4
Scepticism towards AI in mental health
The idea of using AI as a psychologist was met with intense scepticism, with 54 per cent rejecting it. However, a higher percentage of younger respondents saw the potential for AI in providing companionship to lonely individuals.
5
More optimism among AI users
Those who had used AI tools were generally more optimistic about AI’s potential, especially in everyday tasks and creative processes. In contrast, non-users were more hesitant, though both groups shared privacy concerns.
Machine translations dominate low-resource languages on the web
Researchers from AWS AI Labs and UC Santa Barbara have studied the prevalence of machine translation (MT) across the web.
Key findings
1
Machine translations dominate smaller languages
A significant portion of web content in lower-resource languages, particularly content translated from English, is created via machine translation (MT). This dominance is due to the widespread availability of low-cost MT technologies.
2
Low-quality English content is widely translated
The study suggests that much of the machine-translated content originates from low-quality English text, likely produced to generate advertising revenue. This low-quality content is then translated into multiple languages, reducing the resulting translations' accuracy and fluency.
3
Multi-way parallelism affects translation quality
When content is translated into multiple languages, the quality of these translations decreases. The more languages a text is translated into, the greater the likelihood of inaccuracies, with a marked decline in fluency and an increase in predictable, low-quality content.
4
Bias in the type of translated content
The report also identifies a topical bias in translated content. Short, predictable texts with limited complexity—often from conversational or opinion-based sources—are overrepresented in machine-translated data.
5
Risks for multilingual models
Reliance on machine-translated content poses challenges for training multilingual language models. The low data quality in specific languages increases the risk of errors and hallucinations. This calls for better filtering and detection mechanisms to ensure the quality of web-scraped data used in model training.
AI is reshaping journalism and the public arena
Tow Center for Digital Journalism has published a report investigating the structural implications of AI for the news industry. The report is based on interviews with 134 news workers and 36 experts.
Key findings
1
AI is driving efficiency, but the effects are limited
AI helps automate mundane newsroom tasks like transcription, data analysis, and content recommendations. While this increases efficiency, particularly in repetitive tasks, it has not been a "silver bullet" for journalism. Factors such as AI output unreliability and potential reputational damage limit its widespread application.
2
Growing dependence on platform companies
News organisations increasingly rely on the AI infrastructure and services of tech giants like Google, Amazon, and Microsoft. While more prominent news outlets may develop in-house AI capabilities, most smaller players opt for third-party solutions due to cost constraints. This deepens the industry’s dependency on platform companies.
3
AI risks undercutting journalistic autonomy
AI systems from external providers may introduce biases or errors in journalistic output. News organisations face challenges in maintaining control over editorial decisions as AI systems automate content filtering, personalisation, and even the creation of news stories.
4
Unequal access widening the AI gap
Larger, better-funded news organisations are gaining a competitive edge in AI adoption, while local news outlets, particularly in the Global South, are being left behind. This imbalance may deepen as platform companies increasingly dominate the AI landscape.
5
AI-driven media are reshaping the public arena
AI is not just affecting newsrooms but also transforming how news is consumed. AI-powered search and content recommendation systems alter news visibility, with risks that platform-controlled AI might deprioritise high-quality journalism in favour of more commercially attractive content.
From technological optimism to technological race
The Munich Security Conference published its annual report on global security factors. The report covers a wide range of themes and sectors, and here are only a few key relevant (digital) findings.
Key findings
1
Race for dominance
Technology has shifted from being a driver of global development and growth to a geopolitical race for dominance – especially in artificial intelligence. The strategic importance of technological development means that states actively use export bans to deny rivals access to advanced semiconductors: 'From Moore’s Law to More Laws'.
2
Uncontrolled AI race
The US and China compete for dominance in AI, but it remains to be seen who is leading. The uncontrolled AI race poses a security risk due to the unpredictable use and early implementation of AI models – particularly concerning their use in weapon systems.
3
Tool for autocracies
Autocracies have turned digital infrastructure into a tool for controlling populations and undermining democracies, with China and Russia being prominent actors.
4
Absence of global regulation
Increased technological competition between countries leads to less efficient technological supply chains and the absence of global regulation to manage technological risks. The report highlights the need to balance 'inevitable competition' and 'indispensable cooperation', mainly where a win-win scenario is possible.