Exploring How Young People Navigate the Evolving Online World in the Era of Artificial Intelligence and Misinformation

This study is part of a broader investigation on adolescents and misinformation, as well as their online behaviour in Ireland, using an exploratory sequential mixed methods design. Ethical approval was obtained from Dublin City University before conducting this research. In the first phase, a student focus group was conducted with adolescents in Ireland to understand their perception of misinformation and how they navigate the online world (Feijóo et al., 2024). The focus group guided the design of the questionnaire in the second phase of the study, on the results of which the present report will focus. 

The online questionnaire was designed to assess students' understanding of misinformation, artificial intelligence (AI), and their broader online behaviour. The first questions referred to participants' demographics and their overall smart device usage and social media consumption. The participants were then presented with a series of scales assessing their online self-efficacy, susceptibility to misinformation, identification of fake and real headlines, presentation of their online self, and influencer engagement.

The online survey included items specifically developed for this study based on prior research on the topic in Ireland (Feijóo et al., 2023; NACOS, 2021) to inquire about the age of getting the first smart device, daily use of smart devices, social media, apps and different social media features usage. Furthermore, previously existing scales were included to assess an array of variables and are described below linked to the construct they measure.

Online Safety Self-Efficacy. Participants' self-efficacy regarding online behaviours was measured using an adapted version of the Students’ Self-Efficacy in Online Safety Scale by O’Higgins Norman and colleagues (2023). The scale uses a 5-point response range from "not at all" to "very." The adapted items for the present study focus on participants' confidence in areas such as managing online risks, safeguarding personal data, and handling inappropriate content. The report will describe the items specifically about AI and misinformation in the Findings, but the full adapted scale is available in the Appendix.

Perception of Online Self. The Scale of Perception of Online Self Scale (POSS) developed by Fullwood et al. (2016) was included. Each item is rated on a 5-point Likert scale ranging from "strongly disagree" to "strongly agree." This measure evaluates participants' perceptions of their online selves, with some exemplary items being "I find it easier to communicate in face-to-face contexts" or "I can show my best qualities online".

Influencer Engagement on Social Media. Participants' engagement with social media influencers was assessed using the Influencer Engagement on Social Media (IESM) Scale developed by Levesque and Pons (2023). This scale evaluates behaviours such as following influencers and engaging with their content.

Identifying between fake and real headlines. To assess misinformation susceptibility, a short version of the Misinformation Susceptibility Test (MIST) was adapted for the Irish context, based on Maertens et al. (2023). This test evaluated participants' ability to identify, resist, or engage with misinformation shaped as headlines as they had to indicate whether each headline was fake or real. Half of the presented headlines were real, while the other half were fake.

The participants were recruited through a convenience sampling by contacting Webwise’s school network. The sample was drawn from post-primary students aged 11 to 19 years. Schools’ principals and designated teachers distributed an online survey invitation to parents/guardians to obtain parental consent and participant assent. Parental consent and child assent were obtained before participating in the survey, and 397 parents consented to their children participating in the online study. Those who consented shared the survey link with their child, including the assent form, and invited their child to complete the survey. All adolescents were informed of their option to decline participation in the study and that their responses would be kept confidential and anonymous. Those who did not consent were not allowed to proceed to the survey, as answers to this question were necessary for inclusion in the final sample. Despite the recruitment efforts, a smaller sample than intended started the survey after receiving consent from their parents or guardians (214 adolescents). After data cleaning, 109 responses could be included for further analysis. Regarding demographics in the final sample (n = 109), 47.7% identified as boys, 48.6% as girls, 2.8% as other, and 0.9% preferred not to say (see Figure 1).

Regarding device ownership, participants answered an open-ended question about when they received their first smart digital device. Most participants received their first devices between 9 and 13 years old, with 10 and 12 being the most frequent ages. A few participants reported getting their first phone between 15 and 16 years old, which may reflect individual parental choices or cultural practices.

TABLE 1 AGE GETTING A FIRST MOBILE DEVICE

Figure 3 illustrates the daily usage of smart devices among participants. The finding shows that 43.1% of respondents use these devices frequently throughout the day.

In comparison, 21.1% of students engage with their devices a few times each day,
indicating a more moderate level of daily usage. Additionally, 20.2% of participants reported using their devices only once per day, and 10.1% use them less often, highlighting limited digital interaction for this segment.

Participants indicated a diverse range of online social networking sites and apps they use (see Figure 4). YouTube emerged as the most popular platform, with approximately 61.5% of participants indicating they regularly use it. Following YouTube, WhatsApp was favoured by 49.5% of participants, while TikTok was chosen by 43.1%. Snapchat came next at 40.4%, and Instagram was mentioned by 33.9% of participants. Additionally, participants highlighted other used apps, including Bedtime History, Patreon, Discord, Fortnite, Google, and Netflix.

The data on online self-efficacy reveals a mixed level of awareness and confidence among respondents regarding online safety and digital skills (see appendices for complete statistics). Notably, a significant portion of students demonstrated high self- efficacy in key areas: 40% of respondents expressed high confidence in determining which videos to avoid posting online. In comparison, only 4.6% reported feeling no confidence in this regard (see Appendix 1). Additionally, 43.1% of students know "how to treat others respectfully online," indicating a solid grasp of online etiquette. 

In terms of confidence in misinformation content, participants demonstrate different confidence levels. For instance, most adolescents feel moderately confident in distinguishing jokes or parodies from real stories, with 56.8% rating their confidence as 4 and 25.7% as 3 on the scale. However, there are notable challenges in areas like detecting bot-generated content, where one in five respondents (20.3%) feels very unsure. Similarly, identifying fake online profiles poses difficulties, with 18.4% expressing low confidence. When it comes to recognising deepfake or AI-generated content, while some rated their confidence as 3 (27.5%) or 4 (22%), a smaller group (10.1%) remains rated 0 or 1 about being confident on this matter.

The findings reveal that participants are generally familiar with social media features. For instance, 30.3% are aware of the blocking button but have not used it, while 50.3% have utilised it in the past. Regarding privacy settings, 66.1% reported that they manage these settings themselves.

Table 3 Familiarity with Social Media Features

Table 4 demonstrates AI-based digital tools awareness, with participants being asked to reply to “When using digital media, which of the following features have you come across?”. The findings show that filters (photo or video) are the AI-based tool they are more aware of, with 61.5% of the adolescents reporting having encountered it. Smart and virtual assistants, such as Alexa and Cortana, seem to be quite present as well, with an awareness rate of 59.6%, while automatic spell check demonstrates is reported by 54.1% of respondents. Some adolescents have engaged with image generators and deepfakes, with 35.8% indicating they have encountered them. Additionally, 27.5% of participants indicated coming across algorithms, and 38.5% report encountering personalised feeds on social media.

Table 4 AI-Based Tools and Features

In your own words, please briefly describe your understanding of the term “misinformation”.

Based on adolescents' responses, misinformation primarily refers to incorrect, false, or untrue information. Common examples include "fake news" and "wrong information," demonstrating awareness of the inaccuracies that can mislead individuals. Students also recognise the intentional nature of misinformation, acknowledging that it can be used to manipulate or deceive others. Phrases such as "used to persuade you differently" show how they perceived that such information is designed to mislead. Furthermore, references to misrepresentation indicate that misinformation often involves twisted facts or inaccuracies, where information may be misinterpreted or deliberately distorted.

There is an acknowledgement of the harmful effects of misinformation, with some students pointing out that it can spread purposefully, often for personal gain or to attract attention, negatively impacting political and social discussions. However, not all misinformation is intentional; some responses suggest it can arise from accidental errors or miscommunication, where information is incorrectly conveyed or misunderstood. Lastly, the term "fake" is frequently used to describe misinformation, reinforcing the notion that it is fabricated or unreal. While students provide some simplistic definitions of misinformation, indicating a lack of depth in understanding its complexities, there is a clear trend toward recognising its connection to digital platforms and social media, along with a mixed perception of the intent behind the spread of misinformation.

Can you list as many examples of misinformation you may encounter when using social media as you can think of?

Analysing adolescents' responses regarding examples of misinformation on social media reveals diverse perceptions and patterns, reflecting a complex interaction with online platforms. One of the most frequently mentioned types of misinformation relates to news and current events. Students consistently highlighted issues such as fake news, misleading headlines, and biased reporting. Many noted the sensationalism often present in reporting (e.g., a story about “selling dog meat with a sheep's head”) and the deliberate dissemination of false political information. These responses indicate an awareness of how media can shape public perception.

Manipulation of media content through technology also emerged as a prominent theme, particularly the role of AI and editing tools. Students frequently mentioned deepfakes, AI-generated images, and photoshopped pictures as sources of misinformation, reflecting a growing concern about the credibility of visual content online. This highlights the challenges posed by advancements in media creation technologies.

Rumours and gossip also emerged as pointed categories, especially concerning celebrities, public figures, and social contexts like peer groups. These examples underscore the social nature of misinformation, where individuals often share unverifiable or sensationalised stories for engagement or entertainment. Similarly, scams and fraudulent activities, including fake profiles, phishing attempts, and misleading advertisements, were another common concern. Many students identified these as direct threats, particularly forms that target users’ financial security or personal data.

While health and medical misinformation were mentioned less frequently, they remain an area of concern for some participants. Examples include diet myths, misleading health advice, and unverified medical claims. Identity-related misinformation was another frequent issue, such as fake accounts and impersonation. Two respondents mentioned demographic inaccuracies (e.g., “age, weight, size” and "gender and age") as forms of misinformation, viewing them as misleading representations of personal information. In addition, concerns were also expressed about the authenticity of online personas and the potential for these false identities to mislead or exploit others (e.g. fake news impersonation).

Educational misinformation emerged as a notable theme, with students pointing out poorly researched or misleading content presented as fact. Examples include viral “life hacks” and questionable educational videos, such as exaggerated claims about products (e.g., “mascara that will change your life”) or information without any actual content.

Additionally, some mentioned “news” or “news headline” as examples of misinformation, indicating that there is a lack of trust in news for some. 

In your own words, describe your understanding of the term “Artificial Intelligence (AI)”. 

The analysis of student responses about their understanding of Artificial Intelligence (AI) reveals a range of interpretations, from basic concepts to more advanced insights, reflecting different levels of familiarity with the subject. Many students define AI as technology designed to simulate human intelligence, often referencing robots, computers, or bots that can perform tasks typically associated with humans. These tasks are usually described as generating responses, solving problems, or mimicking human behavior. This fundamental understanding highlights the widespread perception of AI as an entity with human-like capabilities.

Some students emphasise AI as a tool for convenience, describing it as technology that can "solve problems," "help achieve goals," or "make things easier." Specific examples include AI applications like autocorrect, image or text generation, and creating videos or accounts online. These responses demonstrate an awareness of AI as a practical, problem-solving tool integrated into everyday life.

Others associate AI with advanced computational capabilities, emphasising its ability to gather information, process data, and make decisions based on large datasets. Several responses focus on the machine learning aspect of AI, highlighting its capacity to learn from inputs, adapt, and apply knowledge in different contexts. For example, some students describe AI as a "computer that thinks like a human with all the information in the world" or "very intelligent, capable of solving many problems automatically."

Interestingly, a few responses reflect misconceptions or incomplete understandings. Some students perceive AI as "not real," "fake," or "not smart," suggesting scepticism or limited engagement with the technology. Others express concerns about potential misuse, such as AI being used to manipulate appearances or create inappropriate content. These responses highlight ethical implications and the potential for harm associated with AI (e.g “It can generate images and text that is fake. Homework helper. I don't use it, but others in my class do. People are lazy and don't want to bother.”). Additionally, there is a tendency to humanise AI, frequently mentioning robots and
"thinking computers." This reflects a common cultural framing of AI as resembling human intelligence or behaviour (e.g,, “A computer who thinks like me”), even though many students acknowledge its non-human nature. Fewer students demonstrated a deeper understanding of AI's technical aspects, referencing databases, programming, and machine learning explicitly

List as many examples as you can of Artificial Intelligence (AI) you may encounter when using social media.

Participants demonstrate varying levels of understanding regarding the use of Artificial Intelligence (AI) on social media. Many students identified AI-driven technologies embedded in popular platforms, such as Snapchat’s My AI, Siri, Alexa, and ChatGPT. These tools are appreciated for their ability to facilitate personalised interactions, answer questions, and provide recommendations, highlighting their role as virtual assistants in daily life. 

Creative applications of AI were also notable, with students mentioning tools that generate text (e.g., ChatGPT), create images (e.g., Craiyon, Midjourney), and edit videos or photos. These applications demonstrate AI's ability to enhance user-generated content and provide entertainment, including deepfakes and interactive role-playing scenarios involving fictional characters. 

A number of participants also recognised AI's role in curating social media feeds, targeting advertisements, and moderating content. This indicate that respondents are familiar that these algorithms influence user engagement by presenting tailored content, although their underlying mechanisms often go unnoticed.

From the total sample, 23% responded regarding influencers they follow, mentioning a wide range of influencers and celebrities. The findings indicate that the entertainment industry plays a significant role in shaping participants' interests, including music (e.g., Tylor Swifts, Beyonce), gaming (e.g., Minecraft YouTubers), and digital entertainment (e.g., MrBeast).

Interestingly, one participant mentioned me in response to popular influence, and another brought up a Chinese name for a favourite influencer (科尔⽐ rose). Instead of naming a specific influencer, some participants mentioned the names of sports teams. This indicates that influencers may hold a collective meaning for some adolescents rather than referring to an individual figure. Additionally, some respondents mentioned they do not follow particular influencers. One also mentioned, “Funny videos, teenage make-up and fashion, TikTok dances. No one in particular.” This suggests that for some teenagers, influencers are not a considerable part of their online activities.

Participants’ perceptions and interactions with influencers were examined across
dimensions of self-concept, emotional attachment, content consumption, and content creation (see Table 5).

In terms of engagement with influencer content, a significant percentage of respondents interact with it. For example, 32.1% of respondents "agree" and 6.4% "totally agree" that they view influencers' photos. For reading influencers' posts, 35.8% "agree" and 5.5% "totally agree." Engagement is highest with video content, where 45.9% "agree" and 6.4% "totally agree" that they watch influencers' videos. However, only 14.7% of users agree, and 5.5% totally agree that they comment on influencer posts. Additionally, creating stories about influencers is quite rare, as only 7.3% of users agree, and 4.6% totally agree with this behaviour.

These findings highlight that while participants frequently consume influencer content, content creation related to influencers is significantly lower. The survey indicates that adolescents have limited reliance on influencers in their daily lives. Only 1.8% of respondents believe their days would be significantly affected by the absence of the influencers they follow. In contrast, a substantial portion of adolescents —23.9% and 21.1%, respectively—disagree or strongly disagree with the notion that influencers play a crucial role in their routines. These suggest that the majority of adolescents do not believe that they depend heavily on influencers for their everyday experiences. However, some participants demonstrate moderate emotional attachment to social media influencers, as reflected in their concerns when influencers are inactive. Specifically, 15.6% of adolescents agree and 4.6% totally agree that they feel worried if a favoured influencer has not posted for a while, suggesting a degree of dependency on their activity.

Table 5 Influencer Engagement

Adolescents' perceptions of their online self were measured by the Presentation of the Online Self Scale (POSS). To assess the participants' views, we asked the following question: "For the items listed below, please select the answer that best describes how you feel about yourself in the online world." (POSS; Fullwood et al., 2016). Please see responses to all items for this scale below in Table 6.

Table 6 Presentation of Online Self

Table 6 illustrates how adolescents perceive their online identities and actions. Notably, 31.2% of participants agree that the internet allows them to express themselves, while 33.9% feel neutral. This suggests that while many individuals use the internet for self expression, some remain undecided or believe that it depends on the situation. One notable finding is that almost 40% of respondents reported feeling like the same person online as they are in the real world, indicating a preference for authenticity in the online platform. In terms of the item “I regularly use different personas (roles/characters) online”, collectively, 54.1% disagreed or strongly disagreed, further demonstrating a willingness to remain authentic to themselves online. Interestingly, many adolescents feel more comfortable with face-to-face conversations. About 31.2% agree, while 17.4% strongly agree that in-person discussions are easier than online conversations, reflecting a comfort with traditional, direct communication.

Table 7 demonstrates the correct answer percentages of those who answered questions about spotting real or fake headlines. Participants showed varying awareness of real versus fake news, where eight headlines were presented: four real and four fake headlines. The data reveals a mixed ability among respondents to correctly identify real and fake news, with accuracy varying across different topics. Notably, the highest correct identification rate was observed for the fake headline about Wi-Fi causing brain damage in Irish teens, with 51.4% of respondents recognizing it as fake. This result suggests that respondents were sceptical of sensational health and technology claims. Similarly, 50.5% of participants correctly identified the real news about Instagram launching a "Restrict" feature to combat bullying, indicating some familiarity with recent advancements in technology aimed at promoting social well-being.

On the other hand, the lowest correct identification rate (27.5%) occurred with the fake news about a gaming console being linked to increased violence and addiction. This may indicate that respondents are more likely to believe fake stories that align with existing societal concerns, such as the negative effects of video games. Similarly, only 30.3% correctly identified the story about young candidates in Irish elections, highlighting a potential lack of engagement or interest in political topics, which might also explain the relatively high rate of incorrect and missing responses in this category.

The difficulty in differentiating real and fake news is evident in categories such as "Irish influencers promoting harmful diet products," where only 37.6% of respondents correctly identified it as fake. This result might reflect the likelihood of such stories in today’s influencer-driven culture, where the promotion of questionable products is not uncommon. Similarly, the real news about social media's link to depression among teenage girls had a correct identification rate of 50.5%, indicating some awareness of this widely discussed issue. Overall, the data highlights the variability in the ability to discern real from fake news, with respondents showing greater scepticism towards health and safety-related claims compared to technology and political stories.

Table 7 Identify Between Fake and Real Headlines