Consumption of Soft and Hard News on the Večernji.hr News Website and Readers’ Interests: The Possibility of Using Artificial Intelligence in the Production of Diverse Media Content

Abstract

The aim of this paper is to clarify the difference between soft and hard news and to explore consumer preferences so that newspapers can adapt their artificial intelligence (AI) tools accordingly. The work focuses on the analysis of the results from a focus group discussion held on 26 March 2024, on the reading habits among readers of the news website Večernji.hr. The analysis shows that readers are not fully aware of their reading habits, often overestimating their interest in hard news while underestimating their interest in miscellaneous entertaining content in the media commonly referred to as soft news. In order to verify their statements, a content analysis of the 50 most-read articles on that news website from 25 to 29 March 2024 was conducted, also from the perspective of hard and soft news, so that this data can be compared with the results of the focus group discussion. The analysis of article readership confirmed and further emphasized the readers’ interest in the miscellanea. These findings have been contextualized within previous experiences of using artificial intelligence in the media, which show that AI tools are highly suitable for informative genres based on service data—typically classified as hard news, and are also compatible with some types of soft news. The great interest of readers in soft news suggests that significant effort should be put in its production and that it should not be considered an unimportant supplement to hard news. Artificial intelligence tools are less suitable for creating miscellanea, but they can be helpful in analyzing trends and detecting events. In these areas collaboration between humans and machines is essential, as only a journalist can accurately understand the human dimension and social context, which is the necessary framework for producing soft news.

1. Introduction

This paper presents the results of the focus group discussion held on 26 March 2024, on the reading habits among visitors to the news website Večernji.hr, in order to examine how much they are interested in hard news versus soft news. After that, an analysis of the content of the most-read articles on that website from 25 to 29 March 2024 was conducted, also from the perspective of hard and soft news. The results were then processed and presented statistically, so that the reading data could be compared with the results of the focus group discussion. The findings are contextualized within previous experiences with the use of AI in global media, in order to provide a recommendation for the use of AI on the analyzed website.

Hard news delivers objective and informative updates on breaking events in the realm of socially relevant topics such as politics and economics, along with natural disasters, wars, and international relations. Soft news is oriented towards entertainment and leisure, encompassing a variety of texts with informative, advisory, narrative, descriptive, instructional, chronicling, interpretative, and other functions, in genres such as gossip, travelogues, reviews, recipes, horoscopes, and various types of advice. It is more subjective and less time sensitive. Hard news is generally read more than soft news, especially in today’s digital age, but exceptions are possible. It should also be emphasized that there are more soft news stories because their production is not limited by current events. The media, due to public interest, are obliged to produce both hard and soft news, and new technologies can assist them in various ways.

The development of AI has also impacted journalism and led to changes in everyday journalistic practice. The creation and distribution of news are changing daily, and new tools enable much faster generation of various content, smarter headline suggestions, and possible additional features for articles. Texts can even be personalized much more effectively according to the interests of the readers. Previous experiences show that AI is successfully used for producing news based on service information (weather forecasts, sports results, stock market reports, price movement analyses), as there is no personal approach in these processes, and the content is based on objective data. On the other hand, AI can successfully find entertainment content (lifestyle, food, travel, oddities, etc.) on social media owing to its understanding of search trends, thus enabling soft news production. This report demonstrates that soft news is not merely an unimportant addition to “real” journalism, but rather an essential part that deserves full attention. In this context, it is shown that AI has significant potential in creating content and distributing soft news.

2. Literature Review

Computerization in journalism leads to numerous new phenomena, with robotic, algorithmic, and computational modes emerging in the media, as well as the introduction of aspects of augmented reality and AI (Calvo Rubio & Ufarte Ruiz, 2021). Recent data indicates that 23% of editors in the media regularly use AI (Editor’s Note, 2023). In numerous analyses of the importance and ways of using AI (Fischer, 2022; Lima-Santos, 2022; Jones et al., 2022; Parratt-Fernández et al., 2021; Graefe, 2016), two approaches have emerged. According to the first approach, the use of AI in journalism can be considered a technical aid to the journalist (Sarisakaloglu, 2022, p. 310; Bohrmann, 2022; Elmer, 2022). The second approach suggests that AI subjects journalistic work to algorithmization and automation (Koska, 2022) and that algorithms will not only automate certain tasks and collect and manage information instead of humans, but will also take over their roles as writers and communicators (Pérez-Seijo & Vicente, 2022, p. 48; Calvo Rubio & Ufarte Ruiz, 2021).

Recent research on the impact of AI on the quality of journalistic work highlights the need to change traditional criteria used for assessing journalistic quality. Although content automation has numerous possibilities, it is still limited in areas that require human judgment, namely in creating interpretive and emotionally driven content. Research shows that human judgment is indispensable and that existing journalistic work ethics must be combined with new technological tools, meaning that criteria should be viewed from a dual perspective: social and technological (Calvo Rubio & Rojas Torrijos, 2024). The cited authors emphasize that it is necessary to focus on educating journalists who need to be trained to recognize the capabilities and limitations of AI in order to adapt these tools to their own needs. Also, the role of editorial and ethical oversight in ensuring quality should not be overlooked. Therefore, media companies should establish guidelines for the use of AI in their newsrooms, which is not easy and straightforward because the potential of these tools is still not entirely understood.

Although it may seem like we have only recently begun to consider the impact of AI tools on daily journalistic practices and related far-reaching implications for the fundamental values of journalism, we must be aware of the fact that AI is not a complete novelty but rather a continuation of the technological changes brought about by the computer revolution and digitalization (Dodds et al., 2025). So, AI is not a completely new technology to be adopted, but rather a continuation of the technological momentum present for several decades. These tools must be viewed as accelerators of a paradigm shift and prompts for reflection on the fundamental aspects of journalism, as well as the role of journalists within society and existing power dynamics. AI tools have also significantly reshaped the ways work is organized within media houses and given new meaning to the concept of rationalizing news production and distribution (Simon, 2025).

Research on trends and predictions of shifts within journalism includes the relationship between generative artificial intelligence and the transformation of newsrooms as an important element of change (Newman & Cherubini, 2025). It states that major publishers have already invested significant funds in experiments with new technologies and that the initial results are promising. AI tools have proven to be extremely useful in data summarization, headline suggestions, various types of text editing (from correcting spelling errors to fact-checking), adapting content to specific styles, data collection, and translation. The usefulness of AI tools in transforming investigative and data journalism is particularly highlighted, but the authors also emphasize that media outlets must be aware that these are expensive and time-consuming processes. The application of AI tools is still in an experimental phase, and the next major shift could occur in content transformation, which also carries pitfalls and dangers, in terms of creating synthetic and hyper-realistic content that may be “hard to distinguish from the real thing” (Newman & Cherubini, 2025).

It has been shown, however, that the application of AI depends on whether hard or soft news is being produced. Shoemaker and Cohen (2006) categorize news into two types: the news that needs to be published urgently because it becomes outdated easily and the news for which timeliness is not critical. Patterson provides a more detailed differentiation, according to which hard news relates to the coverage of important events involving top leaders, major issues, or significant disruptions in the routines of everyday life, while soft news is more focused on personality and is more practical (Patterson, 2000, p. 3), emphasizing that in soft news, the aspect of presentation is more important, whereas in hard news, the aspect of institutionalism is key. Curran et al. also point to this aspect, seeing the distinction between hard and soft news in whether the news is contextualized and related to the public good or without regard to the broader context or implications for public policy (Curran et al., 2010, pp. 9–10), the former being hard news, the latter soft news. Patterson points out a significant increase in soft news, suggesting that it accounts for approximately 50% of reports (Patterson, 2000, p. 3). This aligns with the results of a large meta-analysis conducted by Reinemann et al., who could neither confirm nor disprove the predominance of soft news over hard news, but provided criteria for its identification: (1) covered topics (dimension of topic), (2) specific aspects of events or emphasized themes (dimension of focus), and (3) the way events or themes are visually and verbally presented (dimension of style) (Reinemann et al., 2011, p. 232).

AI has proven particularly useful in journalism when it comes to composing articles that contain extensive data, especially those that change frequently and rapidly, such as financial reports, sports reports, weather forecasts, and election results, or analyses of price movements and traffic conditions (Meade, 2023). Research conducted through interviews with Dutch and Danish journalists (Cools & Diakopoulos, 2024), who have been using generative AI tools daily for some time, examined the frequency of use of these tools in four primary phases of journalistic work: news gathering, news production, news verification, and news distribution. According to the respondents’ answers, the greatest dangers and doubts lie in the verification phase. AI tools are most suitable for the production and distribution phase, while the data collection phase holds potential that is yet to be conceived and carefully structured. However, AI algorithms can have trouble understanding the context and nuances of events and may miss important details or fail to grasp the significance of certain events (AIContentfy, 2023). The use of AI in journalism thus operates on the boundary of quantity, coherence, and control (Mooshammer, 2022, p. 511), with the main expectations placed on data analysis, multimedia content search, personalized content distribution, machine translation, misinformation detection, and automatic text production (Canavilhas, 2022). However, at present, AI will only attempt to simulate human behavior (Nguyen & Hekman, 2022, p. 439), and this is because it lacks knowledge of specific contexts. As a result, the usefulness of AI is demonstrated in processing large amounts of data (Li et al., 2022) and in contexts where repetitive work is required, which is suitable for hard news.

Among the advantages of AI noted in the literature, it is repeatedly mentioned that recurring narrative structures are the most likely candidates for automation (Carlson, 2015; Dörr, 2016). In their comprehensive review of the advantages and disadvantages of AI, Opdahla et al. (2023) emphasize the construction of narratives, along with the capacity to find large amounts of data, multimodality, transmediality, and the possibility of updates, while the main concern is the reliability of sources. The same authors also emphasize that AI is very useful for event detection, trend monitoring, and social media analysis—which are prerequisites for creating soft news—only if the shortcomings of AI can be overcome, namely the lack of understanding of pragmatic and social aspects. Benes (2017) notes that the growing demand for entertainment and lifestyle content is not incompatible with the increasing interest in hard news—it is not necessary for hard news to weaken in order for soft news to be sought after. Furthermore, as Conboy (2023) argues, a large group of relaxing topics—simply called miscellanea—will attract regular readers more than serious topics of political journalism; from his perspective, soft news is not some secondary addition to important news about socially relevant events, but rather a core element of journalism that deserves maximum attention because readers specifically seek it, even more than political topics. Since AI can be utilized and is already being used to create hard news, it can free journalists from low-quality and repetitive tasks (Noain-Sánchez, 2022, p. 116), allowing media professionals to focus on verifying the social context that AI has identified as relevant events or trends.

In addition to the speed in analyzing large volumes of input data, which allows journalists and editors to respond quickly to current events, AI is exceptionally effective for analyzing information about readership and audience interests. This capability enables the optimization of content according to the obtained parameters, which leads to an increase in reader engagement. In addition to creating and optimizing content, AI is a useful tool in translating and shaping articles, as well as in proofreading and stylistic refinement of texts, which allows journalists and editors greater engagement in the research aspect of their work and creativity. However, as the literature analysis has established, one must be cautious in the application of AI, as such content is generic and devoid of context. Furthermore, since AI generates content based on large amounts of data from unreliable and unverified sources, there is a risk of spreading false information (Forja-Pena et al., 2024). Finally, the biggest challenge of using AI in journalism is fundamentally maintaining journalistic credibility and social responsibility, specifically regarding the investigative and ethical aspects of the profession, as well as its impact on jobs in the media sector (Porlezza, 2023).

Research conducted on the example of Greek journalistic practices in local media (Palla & Kostarella, 2025) revealed a fundamental problem in introducing AI tools into everyday use. Journalists doubt that it is possible to produce high-quality content without human intervention, although they acknowledge that AI tools can save time and free up space for more complex tasks such as investigative journalism and critical analysis. As many other recent studies have also shown (Cools & Diakopoulos, 2024; Zhaxylykbayeva et al., 2025), the most significant obstacles in using AI are fundamental ethical issues and the core professional values of journalism (neutrality and transparency), which should under no circumstances be neglected or diminished at the expense of the immense potential of AI tools in terms of efficiency and scalability.

3. Materials and Methods

This research was conducted using the focus group method, which belongs to the category of qualitative research. The focus group is a special technique of group discussion aimed at gaining a deeper understanding of the phenomenon being studied. It is conducted in a small group of participants who discuss a specific topic, guided by a professional—the moderator (Milas, 2009, p. 589). The focus group includes a group discussion that seeks to explore and understand the attitudes, opinions, motives, habits, feelings, and beliefs of the discussion participants regarding a predefined topic in a deep and layered manner. In the focus group, the interaction among participants, facilitated by the moderator, aims to generate insights into content that is rarely gathered through methods such as observation or questioning, which are one-directional and exclude interaction. The focus group research fulfills the primary objectives of this study: to gather a range of attitudes, opinions, and meanings, and to enhance the quantitative research that follows the completion of the focus group. The audio recording of the discussion and the subsequent analysis of the original unbiased information form the foundation of the credibility of this qualitative research. The research was conducted in collaboration with the Promocija plus agency from Zagreb, with the following research questions:

• What news do participants/respondents read?
• What sections interest them?
• How much time do they dedicate to reading the news?
• Do the participants recognize the authors of the articles?

A guide for the group discussion was created based on the research questions that the group discussion must answer. The guide, designed as a carefully planned scenario, contains topics and specific questions that are the subject of discussion, serving as a tool to help the moderator of the group discussion to direct the conversation. The guide, in principle but not exclusively, suggests the order of topics and questions during the group discussion, while allowing the moderator to adjust the scenario to the dynamics of the conversation. Aside from the topics and questions, the guide also includes an introduction to the discussion, which contains a brief presentation of the subject and purpose of the research; an announcement of the expected duration of the group discussion; a request for consent to video–audio record the discussion for the purpose of subsequent content analysis, with an emphasis on the aspect of participant anonymity; and establishment of basic rules for the functioning of the group discussion. The focus group was held online, via the Zoom application, on 26 March 2024. It lasted 90 min, and its participants were readers of e-publications.

The participants of the online focus group were recruited through the existing pool of part-time external associates of Promocija plus, which consists of approximately four hundred individuals of different age and educational backgrounds from all over Croatia. The primary selection criterion, which was also the fundamental principle of group homogeneity, was the fact that participants were readers of e-publications (

Table 1. Demographic characteristics of the participants and their interests.

PARTICIPANT 1 (P-1)	F, 18–39, UD, Osijek, e-publications: 24 sata, Index.hr, local newspapers (Novi list, Glas Istre, Slobodna Dalmacija, Glas Slavonije)
PARTICIPANT 2 (P-2)	F, 40+, UD, Zagreb, e-publications: Index.hr, Jutarnji list, Net.hr, T-portal, Sportske novosti
PARTICIPANT 3 (P-3)	F, 40+, HS, Sesvete, e-publications: Jutarnji list, Večernji list, 24 sata, Index.hr
PARTICIPANT 4 (P-4)	M, 40+, UD, Rijeka, e-publications: Jutarnji list, 24 sata, Index.hr, local newspapers (Novi list, Glas Istre, Slobodna Dalmacija, Glas Slavonije)
PARTICIPANT 5 (P-5)	M, 40+, UD, Pula, e-publications: Jutarnji list, Večernji list, 24 sata, Sportske novosti, Index.hr, local newspapers (Novi list, Glas Istre, Slobodna Dalmacija, Glas Slavonije)
PARTICIPANT 6 (P-6)	M, 18–39, HS, Dubrovnik, e-publications: Jutarnji list, 24 sata, local newspapers (Novi list, Glas Istre, Slobodna Dalmacija, Glas Slavonije)
PARTICIPANT 7 (P-7)	M, 18–39, HS, Zaprešić, e-publications: Jutarnji list, Večernji list, 24 sata, Sportske novosti, Index.hr

). Within the technical and financial limitations of the entire recruitment process, efforts were made to achieve an equal proportion of male and female participants, participants aged 18 to 39, participants aged 40 and older, as well as to establish a balance between younger and older participants and an equal share of participants with high school and higher education qualifications. Participants in the focus group cannot be considered representative of the target population, which directly affects the limited cognitive value of the conclusions drawn from the information collected through the focus group. The results of the focus group cannot be generalized to the entire population and, as such, cannot provide an answer regarding the quantity of a phenomenon being examined; rather, they yield specific answers to the question “why?”

The structure of online focus group participants in our research was as follows:

• The participants come from different parts of Croatia: Osijek (1), Zagreb (1), Sesvete (1), Rijeka (1), Pula (1), Dubrovnik (1), Zaprešić (1);
• Female (3), Male (4);
• From 18 to 39 years old (3), 40 years old and older (4);
• High school (3), University degree (4).

By selecting participants from different Croatian regions, the authors aimed to ensure regional diversity in order to reduce local bias, which would undoubtedly affect the results. Also, the nearly equal gender distribution allows for balancing gender perspectives. The age and educational structure also provide insights into understanding the topic, considering the diverse profiles of readers. Such a structured sample, considering key demographic variables, meets the criteria of relevance, although it is not statistically representative of the entire population.

After a detailed analysis of the focus group results, it was verified which articles were indeed read the most, specifically in the week when the focus group was held, from 25 to 29 March 2024. The main goal was to determine whether the participant’s claims about reading interests correspond to the list of the most-read articles, with articles published on the e-publication Večernji.hr selected as the sample. For this purpose, quantitative content analysis was chosen as an “objectivized method that allows certain qualitative characteristics of textual or visual material to be expressed in quantitative indicators” (Lamza Posavec, 2006, p. 153). The articles were analyzed according to the defined matrix, and the following section of this chapter presents the results of the analysis by day, with an emphasis on their distribution by categories and quantity.

Večernji list was selected as one of the most popular daily newspapers in the Republic of Croatia (Ministarstvo Kulture i Medija RH, 2022). The print edition of the newspaper (popularly known as “Večernjak”) began publication in 1959. It contains diverse sections, from current political topics to miscellanea, but the focus from the beginning was on politics, sports, and culture. The Republic of Croatia, as the majority owner, privatized this media company in 1998. Since 2000, it has been owned by the Austrian media conglomerate Styria. The internet portal Večernji.hr was launched in early 1999. Initially, it only published a selection of texts from the print edition. Over time, the online edition began to keep pace with technological changes and began to function as a news portal, regularly redesigning its visual appearance and structure, introducing multimedia content, video news, and numerous additional portals (sports, showbiz, medicine, advertisement, etc.). Večernji.hr introduced digital subscriptions and paywalls in 2021. According to research on the audience of informative-political press (Lamza Posavec & Rihtar, 2003), Večernji list is read equally by women and men. Regarding the age structure, almost 30% of readers are 60 years old or older, and it is least read by young people under 39. Almost 50% of readers have a secondary education, and it is least read by highly educated consumers. Also, Večernji list has the most readers in large cities (almost 50%) and rural areas (30%). According to the perception regarding individual standards, almost 50% of Večernji list readers have an average standard of living. Compared to other readers of the information–political press, the audience of Večernji list is characterized by older age, lower education, a lower standard of living, and a rural background. Večernji list is regularly read by almost 40% of its subscribers.

4. Results

4.1. Focus Group

Participants were tasked with reviewing all the news, from top to bottom, offered by the e-publication Večernji.hr, and afterwards, they discussed what they remembered and which news items interested them. In this case, the participants of the online focus group remembered a variety of news stories, with the following sticking out:

• Big Brother winner paralyzed;
• Davor Štern: Croatia needs new energy company as energy sources dwindle
• Row between Plenković and Milanović over Constitution violations;
• Tonino Picula angers SDP leaders;
• New gas prices;
• Who calms Filip Juričić?
• Role of vitamin D in body;
• 100 new species discovered in New Zealand;
• Battle of sexes in Croatia: who earns more?
• Dalić announces team for upcoming match;
• Medications to avoid with coffee;
• You will never complain about slow Internet again.

The following news items were mentioned two or more times: “Lamb prices increase by 126 percent”, “Filipinos arrive in Croatia as Croats emigrate”, “Problems with ATMs since Friday”, “Traffic accidents hotspots “, “Rise in number of migrants in Bosnia and Herzegovina “, “Tomašević in Večernji list TV studio”, “Cars that Grmoja, Plenković, and Milanović drive”, “Mandžukić earns millions”, and “Former Cibona player seriously ill”. From the mentioned topics, it is evident that Večernji.hr is quite extensive and filled with a variety of news stories. However, too much content can inversely affect the amount of news remembered, as one of the participants pointed out: “It’s a big website, so by the time you reach the bottom—a block. You no longer remember those first things” (P-4).

4.1.1. Interesting and Uninteresting Articles

Participants indicated that they would be attracted by the following headlines if they had the opportunity to read some news in more detail: “You will never complain about slow Internet again” (P-1), “Problems with ATMs since Friday” (P-7), “Croatia needs new energy company” (P-5), “Battle of sexes in Croatia: who earns more?” (P-6), and “Medications to avoid with coffee” (P-2). On the contrary, the topics that participants were not interested in at all were the following: political news and Tomislav Tomašević’s appearance in the Večernji list studio (P-1, P-4, P-5) and Večernjak’s Ruža (P-7).

News related to politics and the upcoming parliamentary elections was generally perceived as uninteresting, so one of the participants repeatedly stated that they skipped these topics immediately: “The first three or four articles… politics, politics, politics, politics, I just ‘ignore’, ‘ignore’, ‘ignore’, ‘ignore’” (P-4). Uninteresting topics are also the ones that have been dragging through the news for a longer period, such as the war in Ukraine: “I wouldn’t open up the Russian–Ukrainian games anymore. Every day they fight, so it’s already… the war has been going on for quite a while” (P-6).

Therefore, participants perceive soft news as interesting and worthy of attention, from which it can be concluded that such market interest creates a need for greater production of soft news, thereby opening up space for the application and more frequent use of artificial intelligence tools.

4.1.2. Interesting and Uninteresting Newspaper Sections

Answering the question about favorite newspaper sections the focus group revealed various interests, such as gastronomy, sports, international politics, news from Croatia, health news, pets, recipes, crime news, technology (computers and mobile phones), and so on: “I love this ‘show’… a bit of entertainment from all these politicians… so it’s like… sometimes I like to laugh and watch some video” (P-1). No participant mentioned an interest in politics: “Politics and those serious topics… I just read the headlines. My husband comments on what is important to me” (P-3). Also, the focus group participants found news related to elections and politics uninteresting, which they pointed out throughout the entire discussion. They also pointed out celebrity content as uninteresting (P-2, P-5, P-7): “I don’t read about Nina Badrić crying over her ex-husband” (P-5).

The statements confirm that readers of news websites prefer columns with soft content, which further supports the thesis that reading practices lean towards miscellanea, making it possible to strengthen the role of artificial intelligence in the production of news that is in high demand in the market.

4.1.3. Recognition of Article Authors

When the discussion about the importance of text authors arose, many of the participants immediately stated that they did not notice it. Only one of the participants mentioned that they noticed the authors of texts in sports articles, while several participants stated that they noticed authors, mentioning their names: “I notice them in sports. But honestly, I can’t think of anyone right now… oh, I just remembered, there’s an inflation of those analysts on the war in Ukraine, for example Mr. Galić, and that guy in Jutarnji, Tabak… I’m not just talking about those two, but generally, today there are analysts everywhere. Kotromanović is also good, he is a soldier” (P-5). One participant mentioned local authors: “I have a couple of locals that I know. I don’t click on a lot of links at all because I don’t like reading their topics” (P-6).

The responses in which participants claimed that they mostly did not notice the authors of news articles, nor did they find them particularly important, also support the thesis that AI will likely take over the preparation and production of content in the future. Furthermore, editors and media owners will likely decide to make greater use of these tools based on similar public opinion surveys, while directing their newsroom journalists more toward the contextualization and interpretation of news related to daily politics and other hard news content.

4.1.4. The Contrast Between Hard News and Soft News

The next part of the focus group dealt with the types of news that participants read. When it comes to news content, a division into hard and soft news has been widely accepted. As previously explained, hard news brings updates about breaking events, while soft news is oriented toward reports about people. Soft news encompasses a variety of texts with informative, advisory, narrative, descriptive, instructional, chronicling, interpretive, and other functions, such as travelogues, reviews, recipes, horoscopes, and various tips. Soft news comprises lighter topics aimed at entertaining the reader, while hard news is most often “breaking”, such as accidents and natural disasters, government activities, international events, wars, and so on. The majority of participants agreed that they first read hard news and later soft news. Soft news serves as a source of relaxation and entertainment: “I would also start with the hard news first because those are the stories that are in focus, that are current, and the lighter ones you read when you have more time to relax and then you read what interests you and what doesn’t” (P-6); “I also start with hard news, and then I skim through some headlines that I normally wouldn’t click on” (P-7); “First hard news because I think it’s more important, and then soft news, I don’t know, to brighten my day” (P-2).

However, one of the participants mentioned that they first read soft news: “Soft… some kind of entertainment, horoscope. Just for fun, something lighter to read” (P-1). One participant stated that they read strictly hard news: “Hard. I might be a bit older. And as you get older, you have more and more fears regarding safety and generally, especially in Europe” (P-5). Furthermore, one of the participants stated that he chose hard news first, selecting how much that news impacted life in Croatia and himself: “I would also choose hard news, but again, not everything. Now, in India, 10 people have died in the flood. What can I do about it now? Going to read and burden myself, I don’t really have as much time as it may seem selfish. And not everything is true. When a headline says ‘important, important’, I won’t click on it. Like those that could concern us, Ukraine which is relatively close, yes, I’ll read that” (P-4).

4.1.5. The Time Participants Spend Reading

Most participants agreed that they spent one to two hours a day reading e-publications. One participant stated that they spent four to five hours reading articles. In terms of the time participants spent reading articles on weekdays and weekends, they expressed themselves equally—some participants read more on weekdays, while others read more on weekends. On weekdays, those who read more were the ones who commuted to work using public transport or worked in an office job, so they were constantly in front of a computer. Participants who read more on weekends cited that they were more relaxed and had more free time during the weekends, allowing them to afford the luxury of reading in peace the news that interested them: “When I’m at work, the first thing I do is read the news, and then again towards the end of the workday. And on weekends, I allow myself a little break in reading” (P-1).

An interesting example is a participant who spent an hour and a half in the car reading the news while waiting for their child to finish a leisure activity: “Today, while I was waiting to pick up my child from the activity, it didn’t make sense for me to go home, so I took my laptop, I have a mobile router in the car, and I spent about an hour and a half on the laptop. Considering that it was raining outside, I read in the car” (P-5).

4.1.6. The Impact of Article Length on Reading

Most participants confirmed that the length of an article affected reading. They mostly read shorter articles and paid more attention to them, while they lost concentration on longer articles: “You have this analysis about the war in Ukraine, and it can really be, I don’t know, especially on Večernji and Index… it’s just that a person can’t concentrate on such long articles, so I avoid that” (P-6); “I start reading something like that with interest on my phone, and I read, read, and read. Maybe five minutes, ten minutes, I read and I read. And then I scroll down and see that little bar on the side, that I’m only a third of the way through, that there’s still more down below… that I hadn’t even reached halfway. And then I give up. Well, that will be an hour… and then I just skim through the rest. I can’t, I just can’t read for 45 min, an hour, no matter how interesting the topic is” (P-4).

Two participants practiced a method where, if they did not have time to read the entire article, they sent themselves a link to that article and read it later: “If I have time, I read the article. If not, and if it’s interesting to me, I send myself the link, and then I read it when I have time. If something interests me, I will definitely read it sooner or later” (P-7); “I do the same as P-7. I send myself the link if I don’t have the time to read it all at that moment. Although I’ve noticed that I read fewer and fewer of those long articles because I don’t have the patience to read them for so long” (P-3).

4.1.7. Discussing Newspaper Articles with Friends and Family

Participants often discussed the articles they had read with friends and family: “Yes. About sports, some current political topics, about hiking, if there is any news related to being in nature, ecology” (P-7). Some news was commented on with irony: “I often comment on the corruption of Croatian politicians in an ironic and sarcastic way with my friends. There’s always some joke about it. I think it’s irony out of frustration” (P-6).

Two participants emphasized how their husbands/partners kept them informed: “My husband reads much more than I do, so he sends me a link, I read it, and then we can comment on it… mostly sports, a bit of politics” (P-3); “Similarly, my partner sends me something and we comment on it too. Some ways to save money, and then we comment when he finds a link on some social network or on websites. Like, what would be best for us considering that we plan to buy a bigger apartment… In that sense” (P-1). Generally speaking, all participants agreed that they discussed hard news more often than soft news with friends and family.

4.1.8. Paid Content of E-Publications

None of the focus group participants ever purchased content from an e-publication. One participant mentioned that they bought the print newspaper for one article: “I bought it when something was important to me, just to keep it as a memento, but that is really exceptional, exceptionally rare” (P-2). The second participant agrees that it was worth the money spent: “I really love Balašević and I bought it when he died; there was a big interview with him in Večernji list, two pages long… When I was buying the newspaper, I was shocked at how expensive it was… Of course, I threw away the rest of the newspaper and kept the article for myself. I don’t think that will happen again” (P-2).

Some participants believed it was justified for certain articles in e-publications to be charged: “That’s the law of the market… There isn’t much profit, but maybe I’m wrong… Surely there is a difference in the quality of the article, and those that are paid for certainly do not have any sensational headlines, only to lack content inside” (P-5); “I believe that someone’s effort, someone’s work should be paid for” (P-7). Ultimately, almost all participants agreed that they did not consider any topic in e-publications worth paying for: “If I were interested in something like that, I would go to the newsstand and read the print edition. I wouldn’t pay for the online edition” (P-2).

These responses suggest uncertainty among readers regarding the quality of the articles. Participants believe that the quality is insufficient to warrant payment of a certain fee or subscription. This supports the thesis that the future will bring inevitable changes in terms of the automation of media content production, a greater volume of production, and the need for more detailed trend analysis. Consequently, modern journalism will transform into a fusion of automatically generated content and the human component based on knowledge and the ethics of the profession.

4.1.9. Articles That the Participants Regularly Read

All participants, mostly men but also women, regularly read articles about sports: “I follow sports quite often… Probably every day” (P-2); “Also sports every day” (P-7); “I read news about the HNL every day since it is now, after many years, an exceptionally interesting situation” (P-5). Other topics and sections that the participants mentioned were show business, technology, or what was currently relevant: “I’m reading showbiz as I mentioned” (P-1); “I like technology… more or less” (P-4).

Based on these responses, it is clear that visitors to news websites read articles daily in the fields of sports, entertainment, and technology, which means that journalism today has a constant need for large amounts of up-to-date and timely information. Today, the most consumed topics are those suitable for automatic generation and structuring (large amounts of data and technical specifications, statistics, and results). Traditional newsrooms, with their structure and work organization, cannot meet the new market demands, and in this sense, it is essential to design new systems based on processing large quantities of information, detecting current topics and viral content, and recognizing readers’ interests that will shape the content. In that sense, AI will not replace journalists and editors, who will always need to filter context and subtle meanings, but will serve as a valuable aid in preparing and structuring information.

4.2. Content Analysis—The Most-Read Articles on Večernji.hr

On Monday, 25 March, four of the most-read articles were published in the News and Showbiz sections, indicating equal interest from readers in news from the entertainment world and current events. The most-read articles in the Showbiz include topics related to celebrities and their personal stories, while the News covers a variety of themes, from local stories to international events. However, it should be emphasized that even in the News articles, the ones most read are mainly curiosities and soft news, and it should be noted that in the case of articles labeled PHOTO, they are generally photo stories with very little accompanying text content.

• Tito’s granddaughter stars in beloved domestic soap opera and gets divorced after 14 years (Showbiz);
• PHOTO Do you know which is the longest village in Croatia? “I’m going to the neighbor’s” can mean a few kilometers away (News);
• Former champion of Yugoslavia today lives in severe poverty: ”Everyone has forgotten me, and I am starving” (Sports);
• Osijek resident Danijel reveals how much life in Germany costs: “He was surprised by one rule, whether you have it or not, you still have to pay” (News);
• PHOTO Do you recognize this face? From a refugee to a Croatian legend earning millions of euros (Sports);
• PHOTO Noisy at Nikolina Pišek’s: Sunday cleaning showcases expensive furniture (Showbiz);
• Gabi Novak on her health condition: “I haven’t left bed for a month” (Showbiz);
• Farm winner dies in accident before meeting his son: She has found a new husband (Showbiz);
• Elderly woman complains of pain and soon passes away: Doctors shocked by X-ray, “It had been in her for 56 years” (News);
• PHOTO Farewell for mother (32) killed in crash near Split (News).

On Tuesday, 26 March, the Showbiz section once again had the highest number of most-read articles (4), confirming the ongoing interest of readers in news from the world of entertainment and celebrities. However, in the first place was a lifestyle topic, specifically the horoscope, which some of the focus group respondents also highlighted as the most desirable content on the portals. As for news from Croatia and the world, the story that dominated was about the collapse of a river bridge in Baltimore after it was struck by a container ship, a topic that the participants in the focus groups also frequently highlighted:

• These zodiac signs are the best liars, claim astrologers (Lifestyle);
• Banks warn customers: Important service unavailable on Good Friday (News);
• Farm winner dies in accident before meeting his son: She has found a new husband (Showbiz);
• Look at what Novak Đoković was driving around Belgrade (Sports);
• Gabi Novak on her health condition: “I haven’t left bed for a month” (Showbiz);
• PHOTO Doris Pinčić leaps with joy in her garden: A peek into her oasis of peace (Showbiz);
• Ukraine facing new crisis: “I am young and I want to live, not go to war” (News);
• Nova TV announces major change: beloved Croatian TV show is coming back (Showbiz);
• Bridge collapses into river after being struck by container ship (News);
• Have you noticed Modrić’s back? Everyone is talking about unusual circles; here’s what it’s about (Sports).

As in previous days, on Wednesday, 27 March, the Showbiz section once again had the highest number of most-read articles, with as many as five out of ten. It should also be emphasized that the Sports section, which had several articles among the most read on each of the analyzed days, led with topics that were not directly related to sports results, but rather to famous athletes, in this case, tennis players Goran Ivanišević and Novak Đoković. When it comes to the News section, the most-read topics were mainly related to accidents and crime news, rather than politics, which aligns with the views expressed by the participants in the focus group.

• Ivanišević left without big pay: Here are his salaries and bonuses earned with Đoković (Sports);
• PHOTO Doris Pinčić leaps with joy in her garden: A peek into her oasis of peace (Showbiz);
• Ivanišević speaks out after Đoković sacks him: Croatian legend reacts (Sports);
• Actress from “Kumovi” on tough times: “He embraced my loneliness and pain, and 30 years later, I still return to him” (Showbiz);
• PHOTO Peek into Nina Badrić’s luxurious villa: A beautiful oasis of peace in Hvar (Showbiz);
• Wife of Vatreni’s new hero captivates with her beauty: Check out what the British “The Sun” says about her (Showbiz);
• PHOTO First images from crash site: All emergency services on the scene, five helicopters deployed (News);
• Robert and Anica Kovač: Marital bliss for over 20 years after a rocky start (Showbiz);
• Mother of little Danka (2) had her phone seized—surveillance camera footage leaked just two minutes after Danka’s disappearance (News);
• Serbian media: New lead in missing Danka case halts search operation (News).

On Thursday, 28 March, there was a change at the top of the most-read articles on Večernji.hr. That day, the audience was primarily reading articles from the News section, especially those related to the disappearance of a little girl called Danka Ilić, in Banjsko Polje in the Republic of Serbia. Although she had disappeared two days earlier, the news about her became increasingly read as time went on, and she was not found. It should also be emphasized that for the first time in the analyzed period, an article from the section City of Zagreb was on the list of the most read, and it was at the very top. However, it was once again just a photo story with a light theme, and in the eighth place among the most read was a fashion topic about clothing from the 1960s:

• PHOTO Lines at Zagreb’s fish markets: Check out prices of sea bream, hake, and sea bass (Zagreb);
• Ivanišević speaks out after Đoković sacks him: Croatian legend reacts (Sports);
• Wind and thunderstorm alert in Croatia: Temperatures near 30 °C this weekend (News);
• Check your supplies: Popular chocolate bar withdrawn from market (News);
• Mother of little Danka (2) had her phone seized—surveillance camera footage leaked just two minutes after Danka’s disappearance (News);
• Serbian media: New lead in little Danka’s disappearance halts search (News);
• Police issue statement on missing Danka (2): Finding her is absolute priority (News);
• 1960s Fashion in Zagreb: Old photograph of two ladies captivates at first glance (Fashion);
• “Kumovi” actress on tough times: “He embraced my loneliness and pain, and 30 years later, I still return to him” (Showbiz);
• Waiting 14 years for pension: “They are probably waiting for me to die to avoid paying” (News).

Stories from previous days remained relevant on Friday, March 29. In the News section, the disappearance of Danka Ilić continued to be followed, while in Sports, the dispute between tennis coach Goran Ivanišević and tennis player Novak Đoković was covered. When it comes to further thematic analysis of the articles, it should be noted that for the first time in the five analyzed days, one article addressing a political topic made it onto the list of the most read. It is not original content from Večernji.hr, but rather a transcript of an interview with the President of the Republic of Croatia, Zoran Milanović, for the television company RTL. On that day, two lifestyle topics with advice for readers were also emphasized:

• Đoković finally reveals why he parted ways with Ivanišević (Sports);
• Do you want to lose weight by summer? Follow these 7 tips (Lifestyle);
• Robert and Anica Kovač: Marital bliss for over 20 years after a rocky start (Showbiz);
• Day 4 of little Danka’s disappearance: Excavators clear search area (News);
• Major Croatian highway closing for a month: Here’s what’s happening (News);
• Interview with Milanović: Is he still running for Prime Minister and what is his program? (News);
• Good Friday Today: Do you know the customs and why we commemorate it? (Lifestyle);
• Breakup explained: Goran looked like a martyr, as arguments got uglier (Sports);
• Once-in-a-lifetime spectacle coming to the sky: Visible to naked eye—a treat ahead (News);
• Inconsistencies in little Danka case: Mother deleted messages and photographed her from different angles (News).

Taking into account the entire analyzed period from 25 to 29 March 2024, the top ten most-read articles by number of views (

Table 2. Most popular articles (views).

Articles	N	%
These zodiac signs are the best liars, claim astrologers	317,935	9.52%
Do you want to lose weight by summer? Follow these 7 tips	196,263	5.88%
Chic and comfortable: The most beautiful spring shoes we can’t wait to wear in warmer days	167,762	5.02%
Ivanišević speaks out after Đoković sacks him: Croatian legend reacts	146,346	4.38%
8 subtle signs that reveal your dog is lonely	145,635	4.36%
Farm winner dies in accident before meeting his son: She has found a new husband	128,773	3.86%
Robert and Anica Kovač: Marital bliss for over 20 years after a rocky start	128,148	3.84%
Sport	109,804	3.29%
The most beautiful and affordable flowerpots; you’ll want them all	109,405	3.28%
Banks warn customers: Important service unavailable on Good Friday	101,514	3.04%

), excluding photo contributions and galleries, were dominated by lifestyle and showbiz articles, followed by sports and crime news. According to this criterion, soft content takes precedence, which is confirmed by the results of the focus group (readers prefer to consume soft news). Such content is suitable for the application of artificial intelligence tools, with mandatory editorial supervision and journalistic editing, in terms of content creation and assessing the future readership of certain topics, or their potential virality.

If photo galleries are included in the analysis, during the analyzed period from 25 to 29 March 2024 (

Table 3. Most popular articles and photo galleries (views).

Articles	N	%
These zodiac signs are the best liars, claim astrologers	317,935	9.52%
Gallery: Tito’s granddaughter stars in beloved series, loses both parents, and gets divorced.	235,374	7.05%
Do you want to lose weight by summer? Follow these 7 tips	196,263	5.88%
PHOTO Architectural gems across Croatia: Bricked-up doors, a road to nowhere, a broken crosswalk…	186,031	5.57%
PHOTO Lines at Zagreb’s fish markets: Check out prices of sea bream, hake, and sea bass…	180,905	5.42%
Gallery: Milanović’s Life in Pictures—From Childhood to Political Career	174,817	5.23%
Chic and comfortable: The most beautiful spring shoes we can’t wait to wear in warmer days	167,762	5.02%
Gallery: PHOTO Farewell for mother (32) killed in crash near Split	156,508	4.69%
PHOTO Do you know which is the longest village in Croatia? “I’m going to the neighbor’s” can mean a few kilometers away	153,363	4.59%
Ivanišević speaks out after Đoković sacks him: Croatian legend reacts	146,346	4.38%

), the top ten most-read articles by number of views were evenly shared between articles and photo galleries, which refers to content marked as PHOTO. Content analysis again confirms the greater popularity of soft news, with eight out of ten articles belonging to lifestyle, showbiz, and fashion, while only two pieces (sports and crime news) can be classified as hard news. This again confirms that such content is more frequently consumed among readers of online editions.

An active user, in this case a reader, is a person who has visited a specific website within a certain period and has performed some action, such as opening at least one article. In the case of visits to the online edition of Večernji list, that volume of visits was measured from 25 to 29 March 2024.

Table 4. Most popular articles and photo galleries (visits of active users).

Articles	N	%
Ivanišević speaks out after Đoković sacks him: Croatian legend reacts	126,899	27.22%
Banks warn customers: Important service unavailable on Good Friday	82,558	17.71%
Farm winner dies in accident before meeting his son: She has found a new husband	71,387	15.31%
These zodiac signs are the best liars, claim astrologers	28,067	6.02%
Sport	21,639	4.64%
PHOTO Do you know which is the longest village in Croatia? “I’m going to the neighbor’s” can mean a few kilometers away	18,001	3.86%
Do you want to lose weight by summer? Follow these 7 tips	16,059	3.44%
PHOTO Lines at Zagreb’s fish markets: Check out prices of sea bream, hake, and sea bass…	14,175	3.04%
PHOTO Architectural gems across Croatia: Bricked-up doors, a road to nowhere, a broken crosswalk…	12,061	2.59%
Gallery: Doris Pinčić Guberović	9951	2.13%

shows the most popular articles and photo galleries in relation to that criterion. The most-read content was sports-related, followed by service information, lifestyle content, and tips, which again confirms the much greater popularity of soft news.

Table 5. Most popular articles and photo galleries (views per active user).

Articles	N	%
Gallery: Milanović’s Life in Pictures—From Childhood to Political Career	68.99	14.38%
PHOTO Noisy at Nikolina Pišek’s: Sunday cleaning showcases expensive furniture	62.5	13.03%
PHOTO Peek into Nina Badrić’s luxurious villa: A beautiful oasis of peace in Hvar	48.65	10.14%
PHOTO Luxurious estates in Croatia: Once visited by great figures—now crumbling into ruin	41.3	8.61%
Gallery: PHOTO Farewell for mother (32) killed in crash near Split	31.36	6.54%
Robert and Anica Kovač: Marital bliss for over 20 years after a rocky start	28.39	5.92%
Gallery: Tito’s granddaughter stars in beloved series, loses both parents, and gets divorced four years ago.	27.31	5.69%
Chic and comfortable: The most beautiful spring shoes we can’t wait to wear in warmer days	23.52	4.90%
Wife of Vatreni’s new hero captivates with her beauty: Check out what the British “The Sun” says about her	16.83	3.51%
8 subtle signs that reveal your dog is lonely	16.45	3.43%

shows data on the total number of views in relation to the number of active users. As many as seven of the most popular contents are photo galleries and photo contributions, mostly about celebrities. Such content is very receptive to the application of artificial intelligence tools, primarily in terms of tracking reading habits and potential content personalization.

Every reader has their own habits of opening content and staying engaged with it. The analysis presented in

Table 6. Most popular articles and photo galleries (average engagement time per active user).

Articles	N	%
Gallery: Milanović’s Life in Pictures—From Childhood to Political Career	201	14.21%
Sport	197	13.93%
PHOTO Luxurious estates in Croatia: Once visited by great figures—now crumbling into ruin.	94	6.65%
Ivanišević speaks out after Đoković sacks him: Croatian legend reacts	85	6.01%
PHOTO Noisy at Nikolina Pišek’s: Sunday cleaning showcases expensive furniture	80	5.66%
Farm winner dies in accident before meeting his son: She has found a new husband	70	4.95%
8 subtle signs that reveal your dog is lonely	66	4.67%
Robert and Anica Kovač: Marital bliss for over 20 years after a rocky start	66	4.67%
PHOTO Architectural gems across Croatia: Bricked-up doors, a road to nowhere, a broken crosswalk…	63	4.46%
PHOTO Peek into Nina Badrić’s luxurious villa: a beautiful oasis of peace in Hvar	57	4.03%

shows how much time individual readers spent on the news website Večernji.hr viewing specific content. According to this criterion, articles and photo galleries are equally the most popular, with the most-read content being sports, lifestyle topics, and celebrities (showbiz).

5. Discussion

The analysis of the statements and comments from the focus group participants shows that they are only partially aware of their reading habits, as they believe they are more interested in hard news and follow such content more closely, while in reality, they demonstrate a significant interest in soft news, with some of them exclusively following such content. However, some participants explicitly state that topics of great social or global importance do not attract them. Among soft news stories, the participants cover a wide range of topics, from health and pets to media spectacles and technology. Even when they mention hard news, such as crime reports, it is very likely that they are drawn to the personal aspect within them and the human interest elements typical of soft news. The conclusion that they are more drawn to soft news than hard news is supported by the statements of focus group participants regarding their lack of appreciation for authorship and their avoidance of longer articles, which indicates that analytical and critical journalism does not attract them. Among the topics the participants particularly enjoy, sports dominates, although it remains unclear whether sports can be classified as soft or hard news due to its significance. It is very indicative that the participants do not want to pay for content under any circumstances, which again points to a preference for diverse entertainment content, i.e., miscellanea—as it is available in large quantities for free on social media.

When analyzing all 50 most popular articles published on Večernji.hr during the analyzed period from 25 to 29 March 2024, it can be concluded that thematically, the articles published in the News section dominate, which includes the regular sub-sections Croatia, Crime Chronicle, and World, as well as Showbiz. In the third place is Sports, followed by Zagreb, Lifestyle, and Fashion, each with a few of the most-read articles during the analyzed period (Figure 1).

Although the News ranks first in the number of most popular articles, a detailed analysis of these texts shows that only one out of twenty-one articles considers the topic of the interior or foreign policy. In most cases, the topics cover accidents and crime news, both in Croatia and around the globe. Content about various curiosities and oddities is also popular, as well as articles on social issues and problems. It is particularly interesting that articles complemented with additional photographs (marked as PHOTO) had a significant number of views, showing the importance of the visual aspect in attracting attention. The textual component of the content is also important, especially clickbait headlines, or lures, which motivate the audience to click despite the feeling of disappointment upon opening, as confirmed by earlier research (Kanižaj et al., 2022). On each of the analyzed days, about half of the articles on average contained at least a few elements of clickbait (Beck et al., 2021).

It can be concluded that soft news predominates among the most popular articles, regardless of the section, and that the audience largely prefers light topics and advice, along with curiosities and oddities, over more serious subjects related to politics and the economy. The only serious topics that have higher views are those related to crime news, which largely aligns with the opinions expressed by the focus group participants. The answers to the research questions are generally as follows: Focus group participants read both hard and soft news, but they value hard news more, which does not mean they read it more. Similarly, political columns are rated higher, yet statements indicate that entertainment content is followed more closely. Reading mostly occurs outside of working hours, which means that reading newspapers is associated with relaxation, indicating an interest in miscellanea, the diverse entertainment content that soft news provides. In line with these observations, there is also a lack of interest in authored and longer articles, which means that investigative and critical journalism is not followed. The analysis of the most-read content on the news website Večernji.hr during that period confirms the analysis of the focus group statements and partially corrects it, indicating a greater popularity of soft news than the participants suggest.

This subtle discrepancy between the responses of focus group participants and the results of article analysis, i.e., between what the audience claims to read and what they actually consume, points to another aspect of online journalism—the necessity of improving the personalization of media content and responding more effectively to the real preferences of readers. Namely, it is evident that the audience is not always aware of their own habits, as their opinions reflect a common belief that investigative topics and hard news on social and political issues deserve more respect, even though the data shows that these topics are consumed much less frequently than entertainment content, soft news. Classic methods of tracking reading habits can no longer keep up with the changes in the market caused by new technologies and the increasing influence of social media, so artificial intelligence tools can also be very useful in this area. Algorithms based on a large amount of input data (what is most read, how much time is spent on specific content, which topics are most commented on and shared) can offer the audience content tailored to individual interests through recommendations. Today’s tools are already widely used for this type of personalization, but they can be adapted to a greater extent even for hard news, especially news based on structured data, such as sports results, financial trends, and weather reports.

6. Conclusions

This paper presents the results of the focus group discussion held on 26 March 2024, on the reading habits among readers of the news website Večernji.hr. The analysis reveals that readers are not fully aware of their reading habits, often overestimating their interest in hard news while underestimating their interest in the miscellanea—entertaining content in the media colloquially referred to as soft news. However, many focus group participants became aware of their disinterest in political topics and admitted that they enjoyed, sometimes exclusively, sports and lifestyle news. To verify their statements, an analysis of the content of the most-read articles on that portal was conducted from 25 to 29 March 2024, also from the perspective of hard and soft news. The articles were processed and presented statistically, so that the results of these two analyses could be compared. They confirmed and even emphasized the readers’ interest in the miscellany, with soft news on certain days being the most read on the portal. This is contextualized within previous experiences of using AI in the media, which show that AI tools are highly suitable for informative genres based on service data—typically categorized as hard news, and are also compatible with soft news, such as compiling horoscopes, which are, by nature, informative and repetitive. However, the great interest of readers in soft news shows that significant effort must be invested in its production, and that it should not be regarded as an unimportant supplement to hard news. AI tools in the production of miscellanea are less suitable, but they can be helpful in trend analysis and event detection, where collaboration between humans and machines is essential. This is because only a journalist can correctly understand the human dimension and social context necessary for soft news. Soft news content is often not trivial; it requires a framework of broader social context and creativity. While artificial intelligence today successfully imitates human reasoning, it cannot fully capture the subtle nuances that a human author expresses in content creation. However, artificial intelligence can be very useful for content personalization, topic suggestion, analysis and prediction of virality, as well as for linguistic and visual processing of content. Yet, today it is difficult to assess in which direction and at what speed the existing tools will develop and the new ones will emerge, given that the data corpus is growing larger and large language models are becoming more precise every day. Algorithms based on deep learning are also becoming increasingly complex, so prediction and fine-tuning for specific content, as well as the creation of content, will be even more efficient and faster in the future.

Of course, it is always important to keep in mind the dual impact of AI tools. On the one hand, they are changing the technological aspect of content creation and distribution processes through increasingly complex deep learning models, and ever-larger language models. They also serve as instruments for analyzing reading habits, which at the same time shape the way news is consumed as personalization, i.e., the increased attractiveness of content, affects reading habits. Therefore, special attention must be paid to critical human oversight of AI tools. Considering data on the readership of various content, future research should therefore examine the significance of personalization for changes in reading habits in greater detail, paying particular attention to what readers think about their own habits and expectations. Future research should also focus on the effects of automation in content creation, specifically changes in editorial work and journalistic competencies, and the shaping of new work structures in the context of human–machine collaboration. In addition to these social and professional elements, future research should address the issue of media sustainability, the economic impact of changes, and the forecasting and evaluation of new business models. Newsrooms should address some of the following questions very soon in order to re-evaluate their own needs for introducing AI tools and restructure their operations accordingly. These include the following questions: What should be the main criterion for selecting content (popularity, social value, or something else)? How should media content be co-created with AI tools to meet audience expectations and media industry criteria? How can personalized content be developed while simultaneously avoiding reader passivity? In what proportion should content creation be based on predicted audience interests? It is therefore reasonable to expect even greater engagement of artificial intelligence in the creation of media content, which also indicates the need for deeper reflection on the ethically responsible collaboration between AI and humans, a key component of online journalism today, as well as the journalism profession in general. Content must be adapted to the market and the expectations of readers, but one should not lose sight of the role of journalism as a mediator and controller of social processes. In this sense, journalistic content must never become generic, superficial, and entirely predictable. Authenticity and creativity must not lose the battle against speed, optimization, and personalization. Therefore, it is essential to continue conducting analyses and market research regarding the relationship between hard and soft news. This should in turn inform editorial and newsroom decisions about the application of new digital tools and artificial intelligence, while upholding journalistic ethics and the quality of the produced content.