Artificial intelligence is now a day-to-day reality in most London workplaces, with 46% of London workers holding roles with responsibilities that GenAI could automate, compared to a UK average of 38%. What was once thought of as an experimental technology is now standard practice across sectors.
This rapid uptake has prompted anxieties about job losses with 26% of people reporting that they fear AI will replace their jobs. Much of the early discourse on this matter centred around notions that AI would ‘steal’ administrative roles, however, recently, the discussion has turned to positions higher up in the hierarchy.
As AI transformation programmes have grown, so has the creation of visible roles to signal commitment, such as AI Ambassadors, Chief AI Officers at department level, and AI Ethics Officers. The risk with some of these is not that the work is unimportant, but that it creates a focal point that lets the harder questions migrate to someone else’s desk.
An AI Ambassador who builds enthusiasm and fields basic questions during early adoption is doing useful work. The problem is that if governance, accountability and decisions about what to delegate to AI remain unaddressed in mainstream leadership, the ambassador function becomes a distraction from the real gaps rather than a solution to them.
This calls for rethinking what leadership and management outside of AI leadership roles will look like, to support AI Ambassadors and Officers in implementing upskilling and workflow management more widely.
How AI is reshaping the role of the manager
AI already supports some elements of what is typically considered within a managers’ remit: allocating tasks to team members, tracking performance and even weighing up the pros and cons of difficult decisions. It can summarise large volumes of information and data, creating efficiencies in tasks that require high cognitive demand or analysis such as reporting, which once absorbed hours or even days.
While the applications are significant, the technology also has limitations. It cannot read a room, navigate interpersonal relationships, pick up on a change in atmosphere or understand the organisation’s culture. It is in these gaps that the role of the human manager has the potential to flourish.
As AI automates tasks, managers will have improved headroom to deliver a higher-level strategic function and more time to invest in people development. These responsibilities demand skills that are uniquely human, such as critical thinking or understanding the nuance and complexity of situations to make sound judgements, with empathy.
This balance can be understood through our twelve core workplace success skills, a framework that identifies the human capabilities most critical for working effectively alongside AI. It treats AI not as a threat to human skills, but as a tool that, when used well, frees people to do more of what they do best. These span cognitive, interpersonal and adaptive skills, underpinned by a sustainability mindset, cyber responsibility, and systems thinking that connect them into an integrated system where each skill reinforces the others.
The orchestration threshold and the hidden governance gap
Many organisations are approaching a critical threshold in AI adoption, and most are underprepared. Research from MIT NANDA found that although UK businesses invested on average £15.9 million on AI in 2025, no measurable impact has been delivered yet, with failures driven not by technology but by gaps in readiness, governance and leadership.
That gap widens as AI moves from generating outputs to coordinating actions. Orchestration marks the shift from AI as a tool to AI as an active decision-maker, where systems execute tasks before human review. In this model, errors do not just occur, they compound. For managers, oversight shifts from checking outputs to designing systems that can be trusted to act autonomously.
Part of the challenge is that “orchestration” means different things across organisations. For engineers, it is technical coordination. For leaders, it is accountability. For governance teams, it remains an unresolved grey area. Responsibility for multi-agent systems is still fragmented, with no clear standards defining who owns outcomes when things go wrong.
The result is a growing accountability gap. Many managers assume governance is already in place, but often it is not. As AI systems take on more responsibility, leadership must define the boundaries, permissions and safeguards that govern their actions.
Human judgement and critical thinking
Instead, what is being reported is the rise of ‘workslop’, where AI is being used to complete tasks without scrutiny of the outputs. As a result, more time is being spent downstream by managers, fixing a churn of low-quality work. This is creating time drains, where time saved by AI is simply being spent elsewhere in the workflow to correct.
A key skill that managers need to adopt and then embed within their teams is the ability to spot errors, “hallucinations” and biases in AI-generated content, alongside an ability to recognise what ‘good’ looks like. This starts with understanding how AI works, something which many users don’t have as a foundation to build technological skills on top of.
Research from Ipsos revealed that developing employees’ ability to test the accuracy and reliability of AI outputs isn’t where training has been prioritised compared to other AI competencies. Yet this is one of the most critical skills needed and is underpinned by a base-level understanding of what AI is, how it generates content and what risks there are to lifting it at face-value. This isn’t just a blocker to unlocking the potential gains it could offer, but presents a reputational risk to businesses, too.
If AI integration is to translate into tangible results, businesses need to be putting in place the right guardrails, clear use cases and training.
What effective AI training looks like
The national picture reflects what is happening inside organisations. Skills England’s research found that while 97% of organisations reported using AI, most remain at an early stage. Adoption is concentrated in awareness (21%) and exploration (19%), with very few reaching integration (7%), or scale (1%). Many employees are learning through trial and error, without a clear or consistent understanding of safe and effective use.
Skills England addresses this through its PRIMES framework, which sets out six principles for effective AI training. Rather than focusing on specific tools, it defines what good training looks like in practice. For managers, it provides a more structured and reliable foundation than the ad hoc approaches many organisations currently rely on.
Leading by example
Attaching AI to employees’ work without properly training them on how to make use of it risks tools being misused or underused. Enter the new requirement for managers to demonstrate strong AI leadership.
Managers need to be setting clear expectations of AI-usage, determining when it is and isn’t appropriate to adopt and developing the system checks to prevent mistakes. Nurturing adaptive teams that can effectively, ethically and safely integrate AI into workflows is likely to become a growing expectation placed on managers.
Adaptive professionals can operate effectively across roles and projects, against a background of constant change. This makes them versatile contributors who can flex across projects, teams and functions, respond to shifting priorities without constant reskilling and apply judgement and accountability as AI use increases.
There’s also going to be a need for managers to demonstrate their ability to lead teams through change and transformation. Managers will adopt more of a coaching role, nurturing teams to experiment with AI responsibly, so team capabilities multiply, helping colleagues to influence change rather than simply being impacted by it.
It will be those managers who lack AI literacy and the ability to develop adaptive teams that will struggle to lead effectively in an AI-enabled workplace.
Managers are the lynchpin
It is managers and senior leaders that are responsible for enforcing governance around what can be delegated to AI, setting clear expectations around the standard of output and implementing systems to rigorously verify or question outputs. Deciding where it genuinely adds value and then knowing how to embed it will become central to the work of a manager.
Put simply, AI won’t replace managers, but it will reshape what they do. The future’s most effective leaders will combine human skills with AI, to enhance their efficiency whilst doubling down on uniquely human strengths such as complex decision making. They will also play an important role in building adaptive teams, to influence adoption and how AI intersects with workflows, in a way that enhances outputs, while creating efficiencies.
In an AI-driven world, managers are the lynchpin, communicating change, influencing adoption and standards of best practice, while increasing their scope to take on high-level strategic roles.
