DELEGATION (YESTERDAY), ORCHESTRATION (TOMORROW)

Decades ago, organisations often achieved success through a relatively simple management concept: delegation. Management set objectives; managers assigned responsibilities; people executed. And success was all about how well things got delegated, monitored, and done. This worked great, as most of the information analysis, decision-making, monitoring and execution relied on people.

Today, artificial intelligence seems to challenge some of these fundamental premises.

In many ways, the approach to introducing AI into business processes continues to be similar to delegation. People try to define which tasks they can automate, accelerate or delegate to systems for the sake of efficiency. But while these objectives remain crucial today, they may not fully capture the extent of disruption that intelligent machines bring.

AI takes organisations from a realm of delegation to a realm of orchestration. As AI takes on an increasing share of analytical, recommendation, monitoring and execution activities, companies need to coordinate a variety of decisions, accountability, governance, and human judgement across more sophisticated and interconnected systems. Therefore, many of the challenges to scaling AI today come down to orchestration rather than technology adoption.

This is not only important but also increasingly relevant, as the next generation of AI-related challenges may be less about delegating certain tasks and more about orchestrating people, intelligent technologies, data, process execution, and governance into coherent systems capable of producing a desired output.

The difference between delegation and orchestration is subtle yet substantial. Delegation implies assigning tasks and responsibilities. Orchestration is the coordination of the efforts of multiple participants towards achieving a common goal. Managers used to assign tasks and monitor progress. But now the leaders are faced with increasingly challenging tasks in coordinating the interactions among people, intelligent systems, information, process execution, and governance.

As a result, AI brings its own challenges for companies, some of which extend far beyond the adoption of a new technology. Unlearning may be one of the least-mentioned yet critical dimensions of AI transformation. Leaders have developed specific capabilities of monitoring, delegation, performance management and operations management over years and even decades of experience. Some of these capabilities became essential to success.

But they evolved primarily in an environment where most of the analysis and execution relied on people. With intelligent systems becoming a major part of such analysis and execution, some premises underlying these capabilities need to be challenged and reconsidered.

This means that organisations may underestimate the extent of change AI brings if they see the problem as limited to training people to use AI systems. Training may help individuals gain relevant skills and understand emerging technologies. But what many leaders actually need to do is question some of the assumptions behind management and develop new skills in an environment where humans and intelligent systems work together to generate value. In a way, leaders are asked to learn not only how to use AI, but how to rethink decades of experience.

This issue becomes apparent in the context of enterprise-wide scaling of AI projects. In recent years, there have been numerous successes with various AI initiatives. Companies have been able to test the use of various intelligent technologies and achieve positive outcomes. But many organisations still struggle to introduce AI into routine operations across all of their processes and locations. And while a lack of technological competence, data, or integration capacity often comes as an explanation for failed projects, it is rarely the full story.

What a pilot demonstrates is that the technology works. But what an enterprise-wide introduction reveals is an organisation's readiness to adopt new ways of working. As initiatives get scaled up and interact with the rest of the company – governance frameworks, risk management practices, operating models, decision-making, compliance and objectives – they become increasingly dependent on organisational design. In other words, moving away from the pilot stage makes organisations face the orchestration challenges.

And the challenges are significant – especially in the domain of decision-making. Traditionally, people gather information, evaluate alternatives, exercise their judgement and make decisions. But intelligent systems can increasingly participate in many decision-making activities as they perform analytics, spot patterns, generate recommendations, monitor progress, and identify risks.

All this doesn't mean that human judgement isn't needed anymore. Instead, the process of decision-making evolves into a collaboration of people and AI systems in different stages of the decision-making process.

Leaders find themselves needing to determine where people need to continue their contribution, where AI can be a helpful assistant and where some activities can be eventually automated and/or delegated to technology. At this point, making decisions is not the only challenge. More important is the creation of decision processes where appropriate decisions get coordinated and supported by governance and technology.

In this light, training programmes alone may prove insufficient. Organisations can educate people in using intelligent technologies. But the challenges may remain related to the orchestration of processes and decision-making within these processes. Governance systems, decision rights, performance management and other key dimensions of operations may need to be adjusted in order for people and AI systems to interact efficiently to support better decision-making and execution.

The objective is not to create an army of remarkably well-technologically literate employees. The goal is to ensure that companies can operate in environments where intelligent technologies play an increasingly crucial role in analysis and execution.