What social media may have changed in Gen Z at work, and what AI could change next.

Gen Z is no longer the generation preparing to enter the workforce.

It is already there.

Its members are employees, team leaders, managers and, increasingly, employers. They are bringing capabilities developed inside the most connected communication environment in history. They are also entering workplaces reporting high stress, low belonging, communication difficulties and growing tension between generations.

It would be convenient to dismiss these differences as attitude problems. It would be equally convenient to blame everything on social media.

Neither explanation is sufficient.

The more useful question is what happens when people shaped by one social environment enter institutions built around another.

Gen Z broadly includes those born from the late 1990s to the early 2010s. Its members passed through adolescence while smartphones and social platforms became part of everyday life. Social approval became public and numerical. Communication became continuous but increasingly editable. Identity could be presented to an audience, measured through reaction and revised in response.

The workplace demands a different set of social capacities. People must respond without always having time to edit. They must interpret tone, body language and silence. They must cooperate with people they would not personally choose, accept delayed recognition, tolerate ambiguity and remain in relationships after disagreement.

Now artificial intelligence is entering the same environment. It can draft the difficult email, summarise the meeting, answer the junior employee’s question, recommend the managerial decision and evaluate performance. It may reduce stress and make work more accessible. It may also remove more of the interaction, effort and uncertainty through which people become capable.

This makes AI more than a productivity decision.

It is a work-design decision, a cultural decision and, increasingly, a workplace-wellbeing decision.

Social media did not create adolescence, but it changed its conditions

Adolescence has always involved identity, comparison, belonging and status. Social media did not create those developmental pressures. It changed their scale, visibility and continuity.

Research describes several features that distinguish social platforms from face-to-face environments: availability, publicness, permanence, visualness and quantifiability, among others. These features can increase the frequency and immediacy of social experiences, amplify their intensity and alter their character. The same environment can create connection for one person and pressure for another. (Nesi, Choukas-Bradley and Prinstein)

Before social media, a difficult social moment was more likely to remain within a particular setting. School finished. People went home. Different groups knew different versions of a person. Embarrassment had some opportunity to fade.

The smartphone weakened those boundaries.

A young person could carry the social environment of school into the bedroom. A mistake could be photographed and circulated. Approval could be counted in likes, views and followers. Comparison was no longer limited to the people physically nearby, but extended to carefully selected representations drawn from an almost unlimited population.

The effects were never uniform. Social media allowed young people to maintain relationships across distance, find communities, learn skills, publish creative work and discover opportunities that their immediate environment could not provide. For an isolated young person, an online community could provide real belonging. For a developing creator, it could provide an audience and useful feedback.

But the structure of the environment still mattered.

Identity became more performative

Young people have always experimented with identity. Platforms made that experimentation more observable and persistent. A developing identity could be presented to an audience, judged immediately and preserved after the person had changed.

That may increase expressive freedom, but it can also increase self-monitoring. When identity is repeatedly evaluated by other people, the question can shift from “Who am I becoming?” to “Which version of me receives approval?”

Social feedback became measurable

Human beings have always been sensitive to acceptance and exclusion. Platforms converted those experiences into a visible stream of numbers and notifications.

This does not mean that every user becomes dependent on approval. It means that comparison becomes unusually easy and the absence of reaction unusually noticeable. Feedback is also unpredictable, encouraging repeated checking for a response that may arrive immediately, later or not at all.

Communication became continuous but controllable

Digital communication can be rapid, but it also allows time to choose words, rewrite a message, leave a conversation unanswered or withdraw from it entirely.

Those affordances can be useful. They can help people who need longer to process information, reduce some forms of social anxiety and allow thoughtful communication across distance.

They do not provide the same practice as a live conversation in which another person’s reaction is immediate and the interaction must be adjusted in real time.

The algorithm became part of the social environment

Recommendation systems do not merely display the world. They select parts of it according to predicted engagement.

That can help a person find an interest or community. It can also repeatedly expose a vulnerable user to narrow or distressing material when their previous behaviour signals attention to it. More broadly, it creates a highly personalised information environment in which people may receive different accounts of the same world.

What did social media actually do to mental health?

The evidence requires restraint.

The US Surgeon General’s advisory on social media and youth mental health cites a longitudinal study in which adolescents reporting more than three hours of social-media use per day had approximately twice the risk of poor mental-health outcomes, including symptoms of depression and anxiety. The advisory also describes benefits involving connection, self-expression, information and support.

The US National Academies’ review reached a more cautious population-level conclusion. Its review did not support the claim that social media caused changes in adolescent health across the population. Many studies identify associations rather than causation, while the direction can operate both ways. Social media may affect a young person’s mental health, while their existing mental health may affect what they seek online and how they use it.

A 2025 study in Nature Human Behaviour analysed data from 3,340 young people in England. It found important differences in how adolescents with and without mental-health conditions experienced social media, including social comparison, feedback, control over time, friendship and self-disclosure. However, the data were collected in 2017 and were cross-sectional. They cannot tell us whether social-media experiences contributed to mental-health conditions or whether those conditions shaped social-media use.

Even the claim that social media has broadly damaged social skills remains uncertain. A 2024 longitudinal study found that social-media use appeared neither to harm nor benefit the development of social skills overall. (Steinsbekk and colleagues)

The defensible conclusion is therefore not that social media created a damaged generation.

It is that social media became part of the environment in which a generation developed, rewarding some behaviours, reducing the need to practise others and interacting differently with each person’s existing vulnerabilities.

Social media is also only one part of the story. Gen Z encountered economic insecurity, changing education and parenting practices, political polarisation and, for many of its younger members, the social disruption of COVID-19. Age effects, historical events and genuine generational effects are difficult to separate.

The generation shaped by the feed enters the workplace

The transition into work brings these environmental differences into contact with organisational life.

The digital environment often allows communication to be edited, delayed or abandoned. Work requires people to explain themselves under pressure, interpret imperfect messages, negotiate competing interests and continue working together after conflict.

The feed rewards immediate reaction. Careers often reward patience and cumulative competence.

Online communities can be selected around shared interests or beliefs. Workplaces bring together people of different ages, values, communication styles and levels of authority.

Digital feedback is visible and frequent. Workplace feedback can be inconsistent, private, delayed or badly delivered.

This does not mean that younger employees are incapable of workplace relationships. It means that some are entering work with extensive experience in digital communication but less practice in particular forms of embodied, unscripted and hierarchical interaction.

COVID-era remote education and early remote work may have widened that practice gap. It also affected the people who were meant to teach them. Many managers were trying to lead distributed teams while learning how to do so themselves.

What organisations are seeing now

The workplace evidence does not prove that social media caused current problems. It does show a cluster of wellbeing and cultural challenges among younger workers that organisations cannot dismiss.

Stress and disconnection

In Deloitte’s 2025 global research, 40 per cent of Gen Z respondents said they felt stressed or anxious all or most of the time. Work was one contributor, alongside long-term financial concerns. (Deloitte)

Australian data also shows that loneliness has become a significant issue among younger people. The Australian Institute of Health and Welfare reports an increase in loneliness among people aged 15 to 24 since 2012, particularly among young women.

The problem follows young people into employment. In an American Psychological Association survey, 45 per cent of workers aged 18 to 25 said they felt lonely while working, significantly more than older employees. (APA)

This can look paradoxical. A generation with continuous access to communication can still feel socially disconnected.

Connection is not measured by the number of messages exchanged. Workplace belonging is built through being known, being useful to other people, sharing experiences, receiving support and contributing to something with others.

That kind of belonging cannot be created through a policy announcement or another communication platform. It is built through repeated human behaviours: listening, emotional regulation, empathy, recovery and the small daily interactions that help people feel part of a team. Breathe Australia’s Workplace Wellbeing Workshops are designed to develop these practical foundations through sessions covering resilience, emotional agility, listening, conflict resolution, feedback and everyday wellbeing habits.

Declining engagement and recognition

Gallup found that Gen Z employee engagement declined sharply in 2024, particularly in foundational areas such as clarity of expectations, recognition, development, access to resources and feeling cared about at work. (Gallup)

By 2025, Gen Z and younger millennial employees were 13 percentage points less likely than in 2020 to strongly agree that a supervisor or someone at work cared about them as a person. Reported opportunities to learn and grow also declined. (Gallup)

These findings are sometimes interpreted as proof that younger workers need constant praise. There is another interpretation.

They may be entering workplaces in which management fundamentals have weakened. The need for clarity, useful feedback, development and human recognition is not unique to Gen Z. A generation accustomed to responsive systems may simply notice their absence more quickly and be less willing to remain when they are missing.

Communication mismatches

Digital communication removes many of the cues people use to interpret intent. A short message can appear efficient to the sender and dismissive to the recipient. Silence can mean concentration, uncertainty, disagreement or simple delay. Without tone, expression and immediate repair, people fill the missing context with inference.

This is not a Gen Z problem alone. Most workplaces now rely heavily on email, messaging platforms and hybrid meetings. The broader communication environment has changed for everyone.

Younger employees may nevertheless be more accustomed to relationships conducted through these channels. That can produce a workplace in which people message colleagues sitting nearby, avoid a phone call that could resolve an issue in two minutes or spend substantial time managing the wording of a simple interaction.

The cost is not merely inefficiency. Small, informal interactions build familiarity and goodwill. Those reserves matter when disagreement occurs. Without them, a neutral message is more easily read as hostile and a minor misunderstanding can become a cultural problem.

Communication capability therefore needs to be developed deliberately. The Breathe Leadership Communication Program focuses on deep listening, cognitive bias, constructive feedback and the daily interactions through which leaders build trust. These are not secondary interpersonal qualities. In digitally mediated workplaces, they are part of the organisation’s operating infrastructure.

Conflict avoidance and escalation

Healthy workplace relationships do not depend on the absence of disagreement. They depend on the capacity to address disagreement without humiliation, aggression or avoidance.

Some digital environments make withdrawal easier. A person can mute, block, unfollow or leave. Work rarely provides the same option. People must raise concerns, hear explanations, negotiate boundaries and continue cooperating.

When employees lack confidence in informal conflict repair, two patterns can emerge. Problems are avoided until resentment grows, or they are escalated into formal processes before anyone has attempted a direct conversation.

Leaders can worsen this by treating every discomfort as a threat to psychological safety. Psychological safety does not mean freedom from challenge, correction or interpersonal tension. It means people can speak, ask questions, admit mistakes and raise concerns without an unreasonable risk of punishment or humiliation.

Safe Work Australia identifies conflict, poor support, lack of role clarity, inadequate recognition, poor organisational justice and harmful behaviour as psychosocial hazards. These hazards frequently interact. A poorly delivered message is more likely to cause harm in a workplace already characterised by uncertainty, low trust and weak management. (Safe Work Australia)

This is why conflict capability belongs inside a wellbeing strategy. Breathe’s Courageous Conversations and Strengths-Based Feedback program helps leaders recognise their default conflict patterns, separate facts from assumptions and practise feedback that combines accountability with respect. Its Psychosocial Safety for Leaders Program extends that work into the wider system, helping leaders recognise psychosocial hazards and maintain both care and challenge during pressure, uncertainty and change.

Different expectations of work

Gen Z has also entered work with more explicit expectations around purpose, flexibility, wellbeing, fairness and values.

Some leaders interpret this as entitlement. Some younger employees interpret older workplace norms as evidence that an organisation does not care about people.

Both interpretations can become caricatures.

Older workers may have accepted harmful practices because they had fewer alternatives or because those practices were normalised. Younger workers can identify genuine problems that previous generations overlooked. At the same time, a strong wellbeing culture cannot promise that work will never be stressful, inconvenient or emotionally uncomfortable.

The task is not to make one generation conform to another. It is to distinguish between avoidable psychosocial harm and the ordinary demands through which competence, trust and resilience develop.

That distinction becomes easier when generational difference is treated as a leadership issue rather than a contest over which generation has the correct attitude. Breathe’s Multi-Generational Leadership Program helps leaders understand different motivators, communication styles and expectations while building practical approaches to feedback, engagement and knowledge transfer across age groups.

The workplace itself has also changed

It would be a mistake to analyse younger workers without analysing the organisations receiving them.

Many employees began their careers through remote onboarding, video meetings and messaging platforms. They missed informal observation, overheard conversations and the small questions that are easier to ask when another person is nearby.

Remote work can increase autonomy and engagement, but it can also reduce social connection. Gallup found that fully remote workers were more likely than hybrid or onsite workers to report loneliness, anger and sadness, even while reporting relatively high engagement. (Gallup)

Physical presence is not enough by itself. An office where everyone wears headphones and attends virtual meetings can be as socially thin as working from home. The value of in-person work comes from interaction, observation, shared experience and access to other people, not merely occupying the same building.

Organisations therefore cannot demand better interpersonal skills while designing work that provides almost no opportunity to develop them.

AI enters an already fragile social system

The next workplace transition is happening before organisations have fully understood the previous one.

Deloitte’s 2026 survey found that nearly three-quarters of Gen Z and millennial respondents were already using AI to some extent in their daily work. Many felt they were adapting faster than their organisations. (Deloitte)

AI can remove administration, improve access to knowledge, accelerate learning and help people communicate. It can also mediate the relationships through which work occurs.

Social media mediated relationships between people.

AI can mediate the thinking, language and decisions inside those relationships.

AI may conceal a capability gap

A polished document no longer proves that its author can explain the underlying reasoning.

An employee can use AI to write an email, prepare a proposal, analyse information or produce a performance plan without developing the ability to do those things independently. The visible output improves while the underlying capability remains uncertain.

A 2025 study of 319 knowledge workers found that greater confidence in generative AI was associated with less critical-thinking engagement, while greater confidence in one’s own ability was associated with more. The study relied on self-report and does not demonstrate cognitive decline, but it exposes an important organisational question: does AI support thought, or allow the appearance of thought without the underlying work? (Microsoft Research and Carnegie Mellon University)

Research in education shows why work design matters. A large field experiment involving secondary-school mathematics found that unrestricted access to a GPT-based assistant improved performance during practice but produced worse unaided performance afterwards. A structured AI tutor preserved much more of the learning benefit. (Proceedings of the National Academy of Sciences)

The lesson is not that people learn nothing with AI.

It is that the design of assistance determines whether AI becomes a tutor or a substitute.

AI may weaken the apprenticeship pathway

Many professional capabilities are developed through low-level tasks.

Junior employees learn by researching, drafting, observing, correcting errors and gradually taking responsibility for more complex judgements. Some of this work is repetitive, but repetition is often how patterns become visible.

If AI removes the formative work, organisations may gain immediate efficiency while weakening the pathway through which experienced people are created.

This creates a long-term risk. Senior people continue to exercise judgement because they developed it before automation. Junior people produce competent-looking work with AI but receive fewer opportunities to build the mental models behind it. When experienced staff retire, the organisation discovers that it automated the training ground along with the task.

The answer is not to preserve meaningless work. It is to identify which forms of effort are developmental and redesign early-career roles so that AI accelerates learning without replacing it.

AI may reduce interpersonal practice even further

AI can help an anxious employee prepare for a difficult conversation. It can suggest questions, identify inflammatory language and provide a private rehearsal environment. These are valuable uses.

But if the employee sends an AI-written message instead of having the conversation, the immediate problem may be handled while the underlying capability remains undeveloped.

The same risk applies to managers. AI can draft feedback, answer routine questions and generate performance summaries. If it allows managers to avoid human interaction, employees may receive more communication while feeling less known.

An organisation can become highly responsive at the interface and emotionally empty underneath it.

AI may intensify work rather than reduce it

The promise of automation is that people will have more time for meaningful work.

That is not the only possible outcome.

When a task becomes faster, organisations may raise output expectations rather than return the saved time to employees. AI can also enable more detailed monitoring, automated task allocation and continuous performance evaluation.

OECD research has found cases in which workplace AI increased work intensity and reduced human-to-human interaction. Some workers reported lower autonomy, while data collection and algorithmic management raised concerns about privacy and fairness. (OECD Employment Outlook)

This changes the psychosocial environment of work. A human manager can make an unfair decision, but an automated system can apply an opaque decision repeatedly and at scale. An employee who does not understand how work is allocated or performance is judged may experience reduced control, weaker procedural fairness and persistent uncertainty.

AI therefore belongs inside the organisation’s psychosocial risk assessment, not only its technology strategy.

AI can also make work better

A credible account must include the benefits.

In OECD surveys, four in five workers using AI said it improved their performance, while three in five said it increased their enjoyment of work. Training and worker consultation were associated with better outcomes. (OECD)

A study of customer-support workers found that generative AI increased productivity most strongly among less experienced and lower-performing employees. It helped diffuse patterns used by more capable workers and was also associated with improved retention. (Brynjolfsson, Li and Raymond)

AI can improve workplace wellbeing when it:

• removes genuinely tedious administration;
• helps people access information without fear of appearing uninformed;
• supports employees with disability or different communication needs;
• provides additional practice and immediate feedback;
• helps managers identify unclear or unnecessarily hostile language;
• makes organisational knowledge easier to find;
• reduces exposure to dangerous or psychologically distressing tasks;
• gives people more time for judgement, creativity and human connection.

These outcomes are not automatic. They depend on what the organisation chooses to optimise.

From attention extraction to cognitive extraction

The social-media era taught us to examine the business model, not only the interface.

Platforms funded by attention tended to develop features that increased engagement: infinite scroll, notifications, public metrics and recommendation systems. The psychological effects could not be separated from the commercial incentive.

AI requires the same scrutiny.

Does a workplace system increase an employee’s capability and independence, or does it make the employee dependent on the system? Does it return time and agency, or does it convert every efficiency gain into a higher performance target? Does it strengthen relationships, or replace opportunities for people to interact?

The danger is not simply that AI will take tasks from people.

It is that organisations will use AI to extract more cognitive output while paying less attention to the human conditions under which sustainable performance occurs.

What psychologically healthy organisations should preserve

The objective is not to recreate the workplace of the past. Older workplaces contained their own failures: rigid hierarchies, poor communication, tolerated bullying, exclusion and silence around mental health.

The objective is to design a workplace in which new technology increases human capability without hollowing out the relationships and experiences on which capability depends.

That requires more than AI literacy. It requires a deliberate investment in the human capabilities that become more valuable as routine production becomes easier: communication, emotional regulation, curiosity, reflection, feedback, conflict repair and the ability to keep learning through change. These are trainable skills, not personality traits that organisations must simply hope their employees possess.

Preserve human onboarding

New employees need more than access to documents and an AI assistant. They need to observe how experienced people think, ask small questions, understand informal norms and develop relationships before a problem occurs.

Teach conflict repair

Do not assume employees know how to raise a concern, receive disagreement, apologise or rebuild trust. These are workplace capabilities that can be taught and practised.

Psychological safety should support honest conversation, not avoidance of it.

This is the practical territory of Courageous Conversations and Strengths-Based Feedback: developing the self-awareness, listening and language required to address difficult issues while protecting trust and accountability.

Protect developmental effort

Require an initial explanation, hypothesis or draft before AI assistance where the task is intended to develop judgement. AI can then challenge, critique or extend the person’s work.

Leaders must also create a culture in which early attempts, correction and uncertainty are treated as part of learning. The Growth Mindset Leadership Program is directly relevant here because it helps leaders build environments where people grow through challenge, feedback and reflection rather than using technology to avoid the experience of not yet knowing.

Keep consequential conversations human

AI may help prepare performance feedback, restructuring communication or a difficult interpersonal conversation. It should not become a shield that allows leaders to avoid responsibility for delivering and discussing consequential decisions.

The Leadership Emotional Agility Program helps leaders recognise reactivity, remain grounded during difficult emotions and communicate with greater clarity under pressure. In an AI-assisted workplace, that capacity helps ensure technology supports the conversation without replacing the leader’s human presence within it.

Measure more than output

An AI system may increase completed tasks while decreasing autonomy, belonging, learning or trust. Organisations should measure those outcomes alongside productivity.

Consult the people affected

Workers are more likely to trust workplace AI when they understand its purpose, limitations and use of data. Consultation also reveals cultural and practical risks that technical testing may miss.

Build places where people are experiencing play

Healthy workplaces need interaction that is not always transactional or measured. Informal conversations, shared meals, mentoring and collaborative problem-solving build the social reserves that help teams withstand pressure and disagreement.

The workplace is where the two revolutions meet

Social media taught us that a technology designed for connection can also become an architecture of comparison.

It taught us that average effects can conceal vulnerable individuals, that commercial incentives become psychological features and that society often begins discussing safeguards after habits are established.

The workplace is now receiving the generation that developed inside that environment. At the same time, organisations are introducing systems that can draft communication, simulate support, automate judgement and measure performance.

This creates a choice.

AI can compensate for weakened capabilities without rebuilding them, or it can provide the structure and feedback through which those capabilities grow.

It can remove reasons for colleagues to speak with one another, or create more time for meaningful human interaction.

It can increase surveillance and work intensity, or return agency by removing low-value tasks.

The difference will not be determined by the technology alone. It will be determined by work design, leadership, incentives and the value organisations place on human development.

Gen Z should not be treated as a warning about what technology does to people. It should be treated as evidence that people adapt to the environments we build around them.

The question for the AI workplace is therefore not simply what the technology can do.

It is what kind of worker, manager and workplace its use will help create.

Questions for leaders

1. Which interpersonal and cognitive capabilities are employees no longer practising because technology performs the task first?
2. Does our workplace provide genuine belonging, or merely continuous communication?
3. Are we teaching people to resolve conflict, or unintentionally rewarding avoidance and escalation?
4. Which early-career tasks are repetitive but still essential to developing judgement?
5. Does AI return time and agency to employees, or simply increase expected output?
6. Which decisions and conversations should remain visibly and accountably human?
7. How will we know whether AI is increasing capability rather than creating dependence?
8. Are productivity gains being assessed alongside autonomy, work intensity, learning, trust and social connection?

The answers will shape more than organisational efficiency.

They will shape the social and psychological character of work.

Research and further reading

• Social Media and Youth Mental Health: The US Surgeon General’s Advisory
• Social Media and Adolescent Health, National Academies of Sciences, Engineering, and Medicine
• Transformation of Adolescent Peer Relations in the Social Media Context
• Social media use in adolescents with and without mental health conditions, Nature Human Behaviour
• Relationships among social media use, social skills and friendship quality
• Social Isolation, Loneliness and Wellbeing, Australian Institute of Health and Welfare
• Psychosocial Hazards, Safe Work Australia
• Gen Zs and Millennials on Mental Wellbeing at Work, Deloitte
• The Remote Work Paradox, Gallup
• The Impact of Generative AI on Critical Thinking, Microsoft Research and Carnegie Mellon University
• Generative AI Without Guardrails Can Harm Learning, PNAS
• Generative AI at Work, National Bureau of Economic Research
• Using AI in the Workplace, OECD
• Artificial Intelligence, Job Quality and Inclusiveness, OECD