AI at Work: The Behavioural Shift We’re Not Talking About

AI is reshaping work faster than most of us can process. Every week brings a new tool that promises to create the efficiency all businesses crave.  

But while most of the conversation still sits around productivity, not enough are talking about the real revolution – human behaviours. AI is changing how we behave and interact with one another today. 

The unseen shift 

The focus we’re seeing on productivity is understandable. Until now organisations have had to tolerate human inefficiency. Decisions took time, feedback loops were slow, communication was messy.  

AI seems to offer a neat fix: cleaner data, faster insight, instant answers. But as machines take over more of the routine and reasoning, something subtler is happening beneath the surface. 

Employees are quietly experimenting with AI (and most likely, not via your approved pathway) to test ideas or refine tone before they speak. Managers and leaders are using it to sound more empathetic, to script conversations, to make sense of mood and morale. Teams are starting to “perform” for the algorithm, crafting messages and behaviours they know will play well with the system that’s now, effectively, always watching. 

For example, an employee might use ChatGPT to polish a tricky email before sending it to their boss, softening the tone or making it sound “more professional.” It changes how they communicate and how they’re perceived in that interaction. 

Or, a manager facing a difficult conversation about an employee’s performance might use an AI tool to help them decide what to say, including phrases thought to sound motivating and compassionate. The conversation feels warmer, but it’s outsourced – created by a tool rather than actual intuition. 

Or, knowing that an internal messaging platform may be able to analyse sentiment, employees start phrasing their messages more positively or avoiding certain keywords. The result is that in an analysis, their team will look more engaged and positive. 

But not everyone is yet able to use these tools in this way – or wants to – which creates an unfair balance amongst teams, and leaves people skeptical at work of the authenticity of what they’re receiving and also of the positive perception of others. 

The danger with all of this is that we end up with workplaces that look efficient but feel artificial; full of polished performance, low trust and potentially further conflict; this time between those that are leveraging AI in flow mode, and those still in fight or flight mode. Those who have got it, and those who don’t get it. 

An MIT study recently declared that 95% of AI pilots are failing. Not due to challenges with the technology itself. It’s the challenge of implementing this new capability into something incredibly complex: the human-led business.  

 Re-framing AI 

If we treat AI carefully as a behavioural partner as well as a performance tool, it can actually deepen how we lead and learn. 

A good example of this is the type of work my team helps businesses with: Handling conflict, giving feedback, managing performance through change. These are the moments most people avoid because they’re hard, emotional, and unpredictable. AI can help here: it can model tone, simulate reactions, help people rehearse tough conversations, and even surface unseen patterns of friction. 

But, ultimately, it can’t take the conversation for us. If managers lean too heavily on the tech, they’ll lose the discomfort that drives empathy and trust. The opportunity is to use AI as a looking glass, not a mask; a way to prepare for the difficult stuff, but not to dodge it. 

Getting intentional about the human side of AI 

To make that shift, organisations need to get intentional about the human side of AI. That means helping managers become sense-makers, not answer-givers; guiding people through competing data and context, not just passing down conclusions. It’s supporting individuals through AI, and using human skills whilst doing it. 

It means helping employees understand and manage their digital shadow: how their tone, language, and digital habits shape the way they’re perceived by algorithms as well as colleagues. 

And it means designing new norms for transparency and trust: when AI is analysing behaviour, how is that data used? Who sees it? What’s the purpose?  

This isn’t a technology challenge. It’s a behavioural, social and cultural one.  

This will require clear, accessible education for all employees, explaining what AI is (and isn’t), with interactive workshops, simple explainer materials, and real-world use cases to help break down complex concepts. 

This will also require a genuine culture of psychological trust, where employees feel comfortable voicing concerns, asking questions, and sharing ideas without fear of judgment or reprisal. This can be achieved through transparent communication about the purpose, scope, and limitations of AI tools, as well as inclusive decision-making that invites employee input from the outset. 

Not every workplace will get there easily. Some will need stricter guardrails than others. But those who do lean into it and flow have the potential to build towards more reflective employees, more human managers, and more psychologically safe teams. 

Used well, AI won’t make workplaces less human – it could make them more so. By offering a private space to experiment, learn and reflect, it can help people build confidence, insight and connection. It can give managers a clearer view of what’s really happening in their teams and free up time for the conversations that no machine can replace.  

The challenge is not whether AI belongs in the workplace, but how thoughtfully we choose to use it – stripping away the noise and giving us the space, insight, and time to be better.