When the Machine Becomes a Colleague and Therapist: What Happens to Us in the Next Five Years?

## A Quiet Shift: From Productivity Tool to Emotional Infrastructure Artificial intelligence is becoming more than an efficiency tool. It is moving into what has previously been the most private human domain: experiences of stress, loneliness, mastery and meaning. In a few years, AI has gone from something “the IT department” tested to an integrated part of everyday work life — and increasingly part of health and care services. When the technology can both reduce workload and increase pace, a new question arises: Will AI make us mentally healthier, or more vulnerable? It is tempting to lean on the big promises. Zoom CEO Eric Yuan, for example, has argued that AI will automate routine tasks and give us a significantly shorter workweek. His point is that when technology takes over repetitive work, time is freed for more value-creating tasks — and maybe more leisure. In Shifter’s coverage of the interview this is tied to a more drastic change in working life, where the workweek could shrink considerably as AI becomes an “invisible colleague” handling much of the administrative burden (Shifter article). That kind of future image is not only about economy and productivity; it is mental health in practice. This article takes a *prediction* angle: What are the most likely consequences for mental health when AI becomes a standard layer in work, communication and digital health — and what should we prepare for? ## Prediction 1: A shorter workweek does not automatically mean less stress A more automated workday can certainly yield mental health benefits. If AI handles meeting notes, summaries, draft reports and planning, the burden of “microtasks” can be reduced. Research on work stress shows that continuous interruptions and high administrative pressure are often a greater strain than the core job itself. On paper, AI can therefore become a stress reducer. But the first prediction is still that a shorter workweek does not necessarily lower stress levels. The reason is simple: historically we tend to fill efficiency gains with new demands. When delivery capacity increases, expectations for speed, availability and quality often increase as well. Thus, the workweek may shrink in hours but tighten in intensity. In an AI-enhanced work culture, the “norm” can shift: faster replies, more frequent deliveries, more parallel projects. This can create a new type of stress — not because people work longer, but because they operate at a sustained high gear. In practice, mental strain can move from time spent to cognitive pressure. **What does this mean for mental health?** - Increased risk of burnout among employees who cannot keep up with AI-assisted tempo. - Greater gaps between “AI-competent” and “AI-uncertain” workers, which can trigger shame, imposter syndrome and anxiety. - More constant evaluation: work can feel like an endless stream of micro-deliveries that never quite finish. **Likely development (2026–2030):** Companies that realize productivity gains will, in the first phase, primarily reinvest them in growth and faster deliveries. Mental health benefits come first when leadership actively decides to "pay out" the gain as real free time, better staffing or fewer targets. ## Prediction 2: AI becomes the new frontline in psychological support Digital health is moving toward low-threshold, always-available services. That makes AI very tempting: it is available 24/7, can provide structured guidance and can scale to many users. We should expect AI increasingly to become the first point of contact for people who feel anxious, stressed or mildly to moderately depressed. This does not mean AI will replace psychologists, but that it will take on a role as *triage* and support between sessions: reminders about sleep, routines, simple cognitive techniques, journaling and reflective questions. For some this can be life-changing, particularly in areas with poor access to health services. But the prediction carries an important caveat: the effect will be uneven. People with good health literacy and stable frameworks can benefit greatly, while those with complex trauma, severe depression or psychosis may receive poorer — or in the worst case, harmful — support. **The positive scenario:** - More people get help earlier, before problems become severe. - Waitlists are relieved because mild cases get better self-support. - Stigma is reduced when people can “talk” without feeling judged. **The risk scenario:** - False reassurance or oversimplified advice in complex situations. - Users forming emotional attachments to a chatbot and withdrawing from human support. - Excessive self-diagnosis based on generic dialogues. **Likely development (2026–2030):** AI support becomes standard in occupational health, insurance and digital health services. Regulation and clinical standards come after adoption, not before. The gap between technological adoption and health-ethical maturity becomes a central conflict. ## Prediction 3: Ethics becomes the new public health policy When AI moves into both workflows and digital health, ethics becomes practical — not philosophical. Questions about privacy, data storage and model training are not just “compliance.” They are about mental security. Many will be willing to share more about feelings, sleep, conflicts and thoughts if they feel AI helps. But mental health data is among the most sensitive types of data. The prediction is we will see a new wave of digital vulnerability: not only because data can leak, but because it can be used for influence. **Ethical breaches that can directly affect mental health:** - Hidden profiling: If an employer gets indirect insight into stress levels or a “risk profile,” it can create fear and self-censorship. - Manipulative design: If an app optimizes for engagement, it can reinforce addiction and rumination. - "Therapeutic" content with commercial aims: Recommendations can be driven by partnerships, not by health needs. In the workplace this becomes particularly sensitive if AI is used for performance measurement. If AI summarizes meetings, analyzes tone or “reads” emotional signals, employees may experience surveillance — even if the intention is efficiency. That produces a psychological side effect: increased social anxiety and reduced psychological safety. **Likely development (2026–2030):** We will see a two-tier norm. The public sector and serious health actors will move toward stricter requirements for transparency and data minimization, while a grey market of “wellness-AI” will grow rapidly. It will become harder for consumers to distinguish quality from quasi-therapy. ## Prediction 4: Relational loneliness increases — even if we talk more AI can become a social crutch: always available, always polite, always interested. It can help with acute loneliness, but at the same time pull us away from the friction of human relationships. The prediction is we will see an increase in *relational loneliness* — the feeling that one lacks genuine reciprocity — even as the total amount of “conversation” rises. In a workday where meetings and communication are increasingly automated (for example when AI summarizes, suggests replies and generates agendas), human presence can become thinner. If we also get a shorter workweek, it may give more time, but not necessarily more community. A shorter week can become more fragmented, with more remote work and greater individual responsibility to “design” one’s life. Here the connection to Zoom CEO’s future image is relevant: If AI truly enables us to work fewer hours, many organizations will continue to use digital platforms and asynchronous communication to get the most out of the reduced time. The result can be more “optimized” collaboration and less informal human contact. **Likely development (2026–2030):** More people will use AI for emotional regulation (for example by “venting” thoughts), but this may paradoxically reduce motivation to seek human support — because AI is easier, faster and less demanding. ## Prediction 5: A new class divide in mental health — based on AI skills Previously digital exclusion was tied to internet access and basic skills. Going forward the gap becomes more subtle: the ability to use AI as a personal assistant, learning coach and self-help support. Those who learn to collaborate with AI will be able to: - Reduce mental load by “parking” tasks with an assistant. - Get better structure in daily life. - Create more space for recovery. Those who do not may experience the opposite: higher demands, more uncertainty and a sense of falling behind. This can have large mental ripple effects: when people feel they lose control over their workday, the risk of stress-related problems increases. The prediction is that AI competency programs become not just a competitive advantage, but a public health measure. **Likely development (2026–2030):** Workplaces that offer training in AI “for everyone,” with special support for those who feel insecure, will see lower stress and higher well-being than workplaces that merely roll out tools and expect employees to keep up. ## Prediction 6: AI will change how we measure and understand mental health Today mental health is often measured through questionnaires, consultations and self-reporting. With AI we can get more continuous signals: sleep (from wearables), activity level, language patterns in journals, and patterns in digital behavior. This can provide earlier warnings about low mood and burnout. But the prediction is that this also creates a new mental burden: *being measured*. When everyday life becomes a dataset, it can trigger performance pressure — even in self-care. Some will become more focused on the “right” sleep score rather than real recovery. Others will feel guilt if they don’t follow recommendations. **Two parallel trends we are likely to see:** 1. **More precise prevention** for those who want it and control their data. 2. **More health anxiety** for those caught in overinterpretation and constant self-monitoring. **Likely development (2026–2030):** We will see a wave of “mental health analytics” in workplaces, intended to prevent sick leave. Without clear ethical frameworks it can, however, be experienced as control and create the opposite effect. ## Prediction 7: The psychologist’s role becomes more “human” — not less A common fear is that AI will replace therapists. A more realistic prediction is that the psychologist’s role becomes more refined: more time for relationship, alliance, nuance and responsibility — and less time on administrative tasks. If AI takes notes, suggests summaries and structures treatment plans, psychologists can gain capacity for the parts of therapy that cannot be automated: tolerating silence, reading the room, handling ambivalence, and holding human responsibility for another person’s life situation. At the same time this will require new skills: - The ability to assess AI support critically. - Knowledge about bias and error sources. - Ethical competence in data handling. **Likely development (2026–2030):** Clinics that use AI properly will offer more accessible and consistent follow-up. Clinics that use AI merely as a cost-cutting measure risk weakening quality and trust. ## What needs to happen for these predictions to land on the positive side? Predictions are not fate. Outcomes are shaped by design choices, leadership and regulation. If AI is to improve mental health, three levels must work together: individual, workplace and society. ### 1) Individual: Use AI as support, not as gospel - Use AI for structure (planning, reflection, summarizing), but be skeptical of diagnostic conclusions. - Look for services that clearly explain data handling. - Schedule “AI-free” times to avoid constant cognitive stimulation. ### 2) Workplace: “Productivity” must include psychological safety If AI can actually shorten the workweek, the gains must be shared in a way that is felt in the body, not just in KPIs. That means: - Realistic delivery expectations even if tools are faster. - Clear boundaries for availability. - Bans or strict limits on emotion- and behavior-analysis of employees. ### 3) Society: Rules for digital health must be readable The most important thing is not just stricter regulation, but *legible* regulation: People must understand what they say yes to. In a market that will be filled with AI-based self-help products, requirements for documentation, transparency and risk management should become a competitive factor. ## Conclusion: Mental health becomes a design problem Over the next five years, artificial intelligence will affect mental health as much through the rhythm of work as through direct health services. The promise of a shorter workweek — as Zoom CEO Eric Yuan has suggested in his vision of AI-driven efficiency — can become a mental health reform, but only if the gains are actually realized as lower burden and more recovery. If not, AI may instead create a new kind of pressure: higher tempo, more self-monitoring and subtler surveillance. The most likely future is therefore twofold: AI will give more people faster support and better structure, while also increasing the risk of loneliness, pressure and ethical breaches in a market that races ahead of regulation. The question is not whether AI will become part of our psychological everyday life — it is who gets to define the terms. In practice, mental health will be a question of technology, yes, but even more about leadership, ethics and which human needs we choose to prioritize. ## Sources - Shifter (unknown author), *“Zoom CEO predicts: AI will drastically cut the workweek”*, Shifter, 2025-11-23