For most of management history, leaders have had one basic assumption they could rely on: their workforce was human.
That assumption is starting to break.
For the first time, organisations are beginning to operate with human employees alongside AI agents capable of producing work, making recommendations, coordinating tasks and, increasingly, taking action.
At DataIQ, we have started calling this Generation T.
The T stands for Terminator. I should clarify that the talk at DataIQ Guild this year was not of killer robots – the label is deliberately playful. But it does speak to the unsettled (and for many unsettling) prospect of operating a genuinely blended human-AI workforce.
The human-in-the-loop problem
The comfortable answer to AI autonomy has been to keep a human in the loop. It sounds reassuring; AI can make recommendations or perform tasks, but a person remains there to check the output and make the final call.
The problem is scale.
A human may be able to review ten AI-generated decisions an hour. What happens when there are hundreds? Thousands? What happens when the AI can produce decisions faster than any person can meaningfully assess them?
At DataIQ Guild this year, this became one of the more interesting debates.
A question posed at Guild was: do we need to move from thinking about human in the loop to human above the loop? This is a distinction that matters. If every automated decision requires a human to approve it, the human eventually becomes the bottleneck. Worse, they may remain nominally responsible while losing the time and context required to exercise real judgement.
The answer cannot simply be to remove people. It is to become much clearer about where human authority truly matters:
- Which decisions should always remain human-owned?
- Where should people be able to intervene?
- What should happen when an agent behaves unexpectedly?
- Are we prepared to switch one off once it has become embedded in an important workflow?
Those are leadership decisions before they are technical ones.
AI changes what expertise looks like
There is another challenge hiding underneath automation. As AI takes on more routine work, it also starts removing some of the work through which people traditionally become good at their jobs. Expertise is rarely created by reading the handbook. Expertise comes from repetition, making mistakes, spotting exceptions and gradually developing the judgement to recognise when something does not look right.
If AI performs the first draft, the first analysis, the first diagnosis and the first recommendation, where does the next generation gain that experience?
That does not mean protecting inefficient work simply because people have always done it, but it does mean asking a question that is easy to overlook: what are we automating away besides the task?
Some of the skills that become more important in a Generation T workforce are therefore:
- critical thinking
- curiosity
- creativity
- the confidence to challenge
- the ability to understand a business outcome rather than simply produce a technical output
AI may make some technical execution dramatically easier, but it does not make judgement less valuable. On the contrary, it may make it more valuable.
Employees can hear what leaders are saying
There is also a very human problem here. Leaders are talking constantly about agents, automation, productivity and doing more with fewer resources. And employees can hear them.
For some, AI is exciting. For others, the message they hear is: “We are building the thing that may replace you.”
That concern should not be dismissed as resistance to change.
The organisations that navigate Generation T well will need to be much more explicit about the role they expect people to play. AI should not simply be something that is done to the workforce. Employees need opportunities to understand where it is being introduced, contribute their knowledge, challenge how work is redesigned and develop the capabilities that will matter as their roles change.
That does not mean promising everybody that their job will remain exactly as it is. Many roles almost certainly will not. It means being honest about the transition while giving people a meaningful part in shaping it.
We are going to have to learn this one
There is no settled operating model for Generation T.
Some organisations will give agents considerable autonomy, while others will maintain much tighter boundaries. Regulation, industry, workforce composition and organisational culture will all shape where those lines are drawn.
The reality is that we will also get some of those decisions wrong.
That is why the leadership challenge now is not to design the perfect human-AI organisation upfront, but to establish the things that need to remain dependable while we learn: clear decision rights, meaningful human authority, mechanisms for challenge and a deliberate approach to how expertise continues to develop.
The agents will improve quickly. The harder question is whether the organisation around them will learn just as fast.
Generation T is one of the leadership challenges explored in DataIQ’s forthcoming report, Deliberate Decisions: How Grounded Leadership is Advancing the Path to Enterprise AI Value, launching on Monday 12th October. It looks at how data and AI leaders are navigating an AI landscape that continues to redefine human judgement, accountability and capability.
The report is based on discussions held at DataIQ Guild 2026, the industry’s premier invite-only think tank for senior data and AI leaders, in partnership with Databricks. Across two days, leaders compare experience, challenge assumptions and explore the judgement calls required to make progress while the conditions around AI continue to change.
