Much of the talk about artificial intelligence (AI) has focused on people being replaced by machines. But how can it help leaders to be more effective, particularly in the public sector where resources are taut? 

Leaders should be safe from replacement for now; AI is making significant inroads into work that is sequential – a lot of what can be reduced to an algorithm can be done by machines. Excluding creativity – writing, originating, producing stuff – computers fed on a diet of other people's product can originate material that's as good as original work. And while it's not necessarily better, it's certainly faster and cheaper. 

But driving up productivity in the public sector may not be reducible to a series of algorithmic steps. People need to be engaged. Maybe the heavy-lifting for this could be done by AI through data analysis – hoovering up and making new sense of what's already going on. Most of the time, leaders are flying blind. 

Effective leadership is about motivating people to do more than they might otherwise be inclined to do – the so-called discretionary effort bump. The route to liberating this added value is in helping staff deliver for the organisation while simultaneously getting what they want. But getting to this juice (understanding individual motivations) is time-hungry and requires deep listening, two commodities often in short supply. 

But what if AI could get a live read on what's going on in the organisation? Leaders armed with live data on how their staff feel in real time and who could see problems emerging long before they appeared might be able to anticipate and deal with them. 

Let's start with the basics. When a chief executive sends out an email, it's possible to find out who opens it. But there will also be data on how long they spent reading it, whether they forwarded it, commented on it and what they did next. Now imagine a challenging email, one about say, cuts. If leaders had a sense of the number of unplanned meetings taking place or cancelled appointments or phone calls, that might be an indicator that things were going well – or not so much. 

What if leaders could go deeper? Technology for looking at how people behave already exists. We all have a data footprint. Supermarkets, for example, use eye-tracking cameras to understand how consumers interact with products. They use this data to design their stores to maximise sales. 

Similar technology could tell leaders how people are paying attention (or not) and behaving in meetings. It could be a factual bulwark against the oppression of female colleagues who have to put up with eye-rolling, being talked over and negative body language that is part of their day-to-day experience. 

Go further still with a read on how people feel right now? Inspectors already informally gather data about culture and mood, intuitively. Data collection focused on how people walk along the corridors – with purpose, ambling, huddling, chatting – could offer an insight into the live culture. Cross reference this data with other known organisational pressures and leaders could gain some insight into what the organisation's capacity for further change might be. The hardest things in leadership are as much to do with timing as skill. 

And all this is before we get anywhere near the actual content of exchanges. Anything that takes place on Teams is now capturable – transcripts are produced in real-time and can be shared immediately afterwards. AI could look at the content of an organisation's meetings weekly, monthly and annually and scan for key words and concepts – what's being talked about, what seems to matter the most, what gets dismissed, how much airtime do key ideas get and so on?

There would be ethical considerations. It could feel very Big Brother and raise questions about who owns our data footprint – and what rights individuals have in relation to it. There's a balance: leaders' legitimate interest in understanding staff and concerns versus creating a culture of fear where passing remarks and gestures could be misconstrued – or weaponised. 

Selling the benefits to staff would be vital, informed consent and active participation essential. But if they could see leaders responding to what they really think, discretionary effort could surge.