The traditional office social fabric is undergoing a profound psychological shift. Rather than walking down the corridor to bounce ideas off a colleague or venting to a teammate over coffee, workers are increasingly turning to AI as their primary workplace confidant, sounding board and creative partner.
According to a comprehensive workplace study by The Human Connection Workplace Index, a staggering 76% of employees now use AI tools for professional advice, whilst 52% use AI specifically for brainstorming, and 37% admit they now turn to AI for “workplace companionship.”
Dr Cindy Friedman-van der Westhuizen, a lecturer in AI Ethics at Stellenbosch Business School Executive Development, says the shift highlights a dramatic change in how professionals view collaboration.
“Relying on AI as your default collaborator is very tempting and many do rely on AI for once highly collaborative tasks such as brainstorming and generating ideas. These might seem like small interactions, but they’re the moments that have always made workplaces work,” she says.
“Every time you ask someone for a second opinion, challenge an idea together or work through a problem, you’re doing more than completing a task. You’re building trust, learning how other people think and strengthening relationships.”
“One of AI’s greatest strengths is that it’s always available, incredibly fast and instant with its reply. AI responds with a complete lack of judgment and doesn’t mind if you ask it to review a document five times, brainstorm basic ideas, explain a concept or help you draft a difficult email.”
While AI delivers speed and efficiency, Dr Friedman Van der Westhuizen cautions that its greatest convenience may also be its greatest limitation. “AI is becoming the first place we go when we need to think something through or need to think through a problem,” says Friedman-van der Westhuizen. “That’s enormously useful. But it’s also replacing the five-minute desk-side conversations where ideas are tested, problems are shared and relationships are built. The typical water cooler moments are now being replaced with a ‘Hi Claude, please help me, I’m stuck’.”
She says that whilst human colleagues bring different lived experiences and uncomfortable but productive pushback, AI is optimised to be helpful. This means working within a user’s framing rather than questioning it.
“AI doesn’t become impatient and it doesn’t disagree based on its own lived experience,” says Dr Friedman-van der Westhuizen. “It functions less like a true collaborator and more like a very sophisticated yes-man. It may help refine your thinking or fill in gaps, but it is unlikely to challenge the premises you start with fundamentally.”
“While work is a site of production, it is a critical hub for human social interaction. Relying on an algorithm to generate ideas, navigate sensitive workplace politics or plan for a difficult conversation, strips away the vital micro-interactions that build corporate cultures.”
“If our first instinct is always to ask AI instead of another person, we may gradually lose the everyday opportunities to practice negotiation, persuasion, constructive disagreement and collaborative problem-solving,” warns Dr Friedman-van der Westhuizen. “Those interactions create social capital inside organisations. If AI starts replacing them, the impact may only become visible years later.”
Dr Friedman-van der Westhuizen says it’s powerful to see how naturally humans anthropomorphize AI. “AI models speak with human fluency and mirror emotional registers, making it incredibly easy for users to mistake AI encounters for genuine human empathy. This powerful illusion is driving employees to navigate sensitive workplace politics and personal anxieties entirely through software. Employees find it easier to trust an unbiased algorithm with their raw, unpolished ideas and corporate anxieties than to risk showing vulnerability to a human colleague or manager.”
However, the risks for organisations are complex. Research has found that 57% of global employees actively hide their AI usage with almost half of employees admitting to using AI in ways that contravene company policies, including uploading sensitive company information into free public AI tools like ChatGPT.
Many rely on AI output without evaluating accuracy (66%) and make mistakes in their work due to AI (56%).
“This complacent use of AI is also an issue of governance and the responsible use of AI. Only 47% of employees say they have received AI training and only 40% say their workplace has a policy or guidance on generative AI use.”
Rather than discouraging AI adoption, Dr Friedman-Van der Westhuizen says organisations should become more intentional about how they use it. “The efficiency gains of AI are real. AI can be beneficial to remove repetitive administrative work, summarise information and improve productivity. However, organisations must recognise that human collaboration is not just about getting work done but about maintaining the social and intellectual skills required to work together.”
She encourages organisations to identify which activities should remain fundamentally human. “There needs to be an awareness that not every task that AI can do is one it should do. Mentoring, collective problem-solving, coaching, conflict resolution and brainstorming should be protected as human spaces where the value lies in the human interaction, not only the output.”
“AI should be used as a tool not as a partner. Organisations need to make it clear that AI can never be a substitute for human engagement. If AI is used for administrative purposes, then in theory that should free up cognitive and social bandwidth for the richer, messier work of human collaboration. To avoid being deskilled by technology, organisations need to invest in developing certain skills intentionally, rather than assuming they will develop organically. This could look like meetings and workshops that are explicitly AI-free, designed to practise negotiation, disagreement and creative problem-solving together.”
Whilst AI policies as a standard protocol cover issues such as data governance, human oversight, bias auditing and employee rights, governance strategies and policy documents need to address the risk of AI replacing human coloration.
Dr Friedman-Van der Westhuizen says for guidance one can look beyond the corporate world. “The African Union’s Continental Artificial Intelligence Strategy , though written for national and continental governance, is built on a human-centred philosophy: that AI should serve and empower people, not the other way around. That principle translates surprisingly well to the workplace, especially in the context of concerns about AI becoming more than a tool and replacing.”
Below are six governance strategies, drawn from the AU framework, adapted for organisations grappling with exactly these concerns:
- Be clear about why you are using AI: Before deploying AI for any workplace function, organisations should ask whether it is solving a genuine problem or simply being adopted because it is available. This matters for collaboration specifically. If AI is introduced into team workflows without a clear rationale, it can displace human interaction by default rather than by design. A “people-first” adoption policy forces that question to be asked explicitly.
- Define what AI should and should not do: Organisations should identify which AI applications are approved, restricted or prohibited and, crucially, why. Given concerns about AI replacing human collaboration, this means actively protecting certain functions as human spaces: mentorship, creative brainstorming, conflict resolution and peer feedback should be identified as areas where human engagement is not just preferable but required.
- Be transparent about AI’s limitations: Employees should know when AI is being used, what it is being used for and, perhaps most importantly, what it cannot do (or is not good at doing). People who understand that AI is optimised to be helpful rather than honest, and that it cannot bring genuine cognitive diversity to a problem, are less likely to treat it as an intellectual equal. Transparency about limitations can help prevent overreliance on AI as a collaborative partner.
- Create a governance structure with real accountability: More than just developing policy documents, organisations should also consider establishing an AI governance committee, or designating specific roles responsible for approving, monitoring, and reviewing AI use over time. This is particularly important as AI capabilities change rapidly because what seems like a reasonable boundary today may need revisiting within months.
- Invest in AI literacy, including its social dimensions: Training programmes should go beyond teaching employees how to use AI tools effectively. They should also address the subtler risks: the tendency to anthropomorphise AI, the erosion of collaborative skills through disuse, and the intellectual costs of working primarily with a system that never disagrees with you. Employees who are genuinely literate about AI are better equipped to use it as a tool without ceding to it the role of colleague.
- Make it safe to raise concerns: Organisations should establish clear processes for reporting harmful AI outputs or ethical concerns about AI use, including concerns about AI displacing human interaction in ways that feel difficult to name or quantify.
