Middle East perspectives | Work in the AI era: Managing a workforce that includes both humans and AI agents

The shift of an organisation’s identity 

Q: What becomes the definition of an employee when humans and AI share the same organisation chart?

The definition of an employee remains human. What is changing is the definition of the workforce.

At G42, we are actively shaping and testing what an AI-native workforce can look like when capability no longer sits only with people. Some capability sits with humans, some increasingly with AI agents, and the real opportunity comes from how intelligently the two work together. This moves us from thinking about workforce simply as headcount to workforce as capability.

But there is an important distinction. AI can analyse, recommend, generate and execute. Humans bring something different: judgment, accountability, creativity, culture and meaning. So while the future organisation chart may include both people and AI agents, responsibility for direction, decisions and consequences must remain human-led.

For Human Capital, this creates a new mandate: architecting how human ingenuity and machine intelligence come together responsibly to create value.

Q: How does an organisation measure trust in a workforce where some contributors are non-human?

Think about how we build trust with a new colleague. We do not give them unlimited responsibility on day one. We observe their judgment, reliability, consistency and how they operate within the organisation’s expectations. We should approach AI with the same discipline.

At G42, we are moving beyond simply deploying AI tools. We are actively testing how agents can be validated, governed and assessed as contributors to enterprise work including defined responsibilities, permissions, performance measures, review periods and escalation points. The principle is straightforward: powerful technology does not automatically deserve trust. Trust has to be earned through evidence.

For AI, that evidence comes through reliability, transparency, measurable performance, governance and human oversight. The more capable AI becomes, the more important these disciplines become.

Workforce architecture and design 

Q: What is the new unit of work when tasks can be decomposed and recomposed between humans and AI agents? 

For decades, the job has been the basic unit around which organisations were designed. AI challenges that assumption. Increasingly, work can be broken down and recomposed between people and intelligent systems. The important unit may therefore become the brief: the human intent, context and problem that directs a partly automated system. AI may execute significant parts of the workflow, but humans determine what problem is worth solving, establish the boundaries, exercise judgment and ultimately own the outcome.

At G42, this is where the conversation is moving from automation to organisational redesign. The objective cannot simply be to use AI to do yesterday’s work faster. It should be to redesign work so that technology removes friction and releases human capacity for higher-value contribution. And that creates perhaps the more important Human Capital question: If AI gives our people time back, what higher-value work are we going to ask them to do with it?

Q: How do you design a workforce where roles are fluid, but accountability remains fixed?

AI will make organisations significantly more fluid. Skills can move between projects, people can contribute beyond traditional job boundaries and AI agents can support multiple workflows simultaneously. That flexibility can create enormous speed. But flexibility without accountability creates risk.

At G42, our principle is that AI may recommend, analyse, draft, coordinate or execute within clearly defined parameters. But accountability for decisions affecting people, culture, ethics, customers or enterprise outcomes must remain clearly human. Intelligence can scale. Accountability cannot be outsourced. That principle becomes even more important as organisations become more AI-native.

Governance, risk and accountability 

Q: What does the organisation’s duty of care become when AI agents make decisions that affect people? 

This is where Human Capital needs to be at the table from the beginning. When AI influences recruitment, performance, learning, mobility or employee experience, these are no longer purely technology decisions. They are decisions about people.

At G42, we believe AI should help us understand people better including their capability, aspirations, potential and opportunities for growth. If technology makes people feel reduced to data points, then we have missed the opportunity.

Our duty of care therefore has to include fairness, transparency, explainability, data responsibility and, importantly, the ability for decisions affecting people to be reviewed and challenged by a human being. The more intelligent our systems become, the more intentional we need to be about their human impact.

Q: How do you govern a workforce where some contributors can self-improve without permission? 

This is one of the most important governance questions in an agentic world. Systems that learn and adapt can create extraordinary value, but they can also drift sometimes, not dramatically, but gradually: a change in prioritisation, behaviour or decision-making that was never intended.

At G42, we are actively testing how AI agents can operate within a structured workforce model: defined roles, permissions, boundaries, review cycles, performance expectations and escalation points. Governance cannot be something we add after deployment as a compliance layer.

It has to be designed into the workforce architecture from day one. And importantly, governance should not be seen as the opposite of speed. Good governance creates the confidence to move faster and scale responsibly.

Performance and productivity 

Q: What is the new definition of productivity when AI operates at non-human scale?

Speed can be seductive. When AI produces in seconds what previously took people hours or days, it is easy to define productivity as more output. But more output is not necessarily more value.

At G42, we are increasingly interested in a different question: does AI help us make better decisions, reduce friction, improve quality, accelerate learning and unlock innovation? If AI simply allows an organisation to produce twice as many reports, that is efficiency. If it allows people to spend less time producing reports and more time solving the problems behind them, that is transformation.

The opportunity is to redirect human energy towards leadership, innovation and strategic outcomes. That, to me, is a much more meaningful definition of productivity.