Six levels of autonomous work: How AI augments, then replaces

Intel The AI revolution is upon us, but it remains extremely hard for business leaders to set a direction and vision and to make plans with any certainty. Still, we can offer some relatively uncontroversial observations regarding current and future capabilities — around which we can start to build a broad picture of this revolution. These include:

• AI is already impressive in its generative and predictive capabilities and is only going to keep getting more so.
• There is a huge amount of investment and excitement in the space, which seems unlikely to abate any time soon.
• CEOs are always on the search to achieve more with less (growth and margin).
• Many jobs — or parts of jobs — are routine, procedural, or algorithmic in nature, and are therefore candidates for reallocating to AI resources. According to H. James Wilson and Paul Daugherty in Harvard Business Review (Sept-Oct 2024), most business functions and more than 40% of all US work activity can be augmented by AI.
• New companies very soon will be AI natives, meaning that they simply will not hire humans in the first place except when they have to. These companies will probably show the rest of us where humans are still valuable and where they’re not, and we’ll follow suit (some faster than others).

On this patchy but still relatively solid ground, we were inspired by “The 6 Levels of Driving Automation” — created by the Society of Automotive Engineers — to develop a framework that reflects this evolution of AI capabilities and how they will affect companies over the next decade or so.

A continuously improving set of AI resources over the next decade will have a two-fold impact on business and the human workforce. Initially, AI will have a broadly augmentative effect, taking over low-value tasks and empowering humans to focus their efforts on more strategic and creative jobs.

But at some stage, likely in five years or so, AI will start to take over entire job roles, starting with the most “procedural” or rules-based jobs. Eventually, it will acquire enough decision-making and orchestration capabilities to take over entire teams and even lines of business.

These two distinct effects, which we’ve labeled an Augmentative phase and a Replacement phase, will likely happen gradually at first, then more quickly. However, the speed and depth of adoption will vary by industry, function, team, and individual.

The 6 Levels of Autonomous Work

King and Afshar – Co Authors

The six levels of autonomous work

What follows is a row-by-row discussion of the chart above.

Level: Each autonomous work level is labeled by number (0-6) and title. The title refers to the amount and complexity of work that AI can do at that level. It is essentially a generic work breakdown, starting with the smallest and simplest chunk of work, namely a Task (level 1). The next level up from a Task is the Sub-Process (level 2), referring to a group of tasks that are typically performed in sequential order to complete a discrete part of a business process, such as ensuring that all relevant information has been collected accurately and completely to open a customer case.

At level 3, AI has the capacity to complete a business process such as taking a customer order, managing a customer case from open to close, and qualifying a lead. At level 4, AI can complete several processes from beginning to end, performing most of the work that would be conventionally allocated by role, like sales representative, marketing specialist, or service agent. We’re focusing here on typical commercial operations but the equivalent will be true in manufacturing and all other types of operations. At level 5, AI or AIs can perform most of the roles associated with any commercial team –including a “manager” and their direct reports — that together execute one or more complex business processes. At level 6, AI can orchestrate the work of multiple teams, functions, and processes, conventionally organized as a business or line of business. Eventually, this will include all small and medium-size businesses, and — in the longer term — large enterprises (although “large” refers purely to business complexity and revenue size, not employee count).

**Phase: **The six levels of autonomous work described above do not represent a linear trajectory for AI. AI will not evolve to more senior roles in an organization in a traditional career progression. Instead, there will be two quite distinct phases in its progression. The first is levels 1-3, which we can describe as the Augmentation phase in which digital assistants will enable and empower human employees to do their best work, and will create new opportunities for them too.

The second is levels 4-6, which is the Replacement phase in which digital agents will take on increasingly large and complex responsibilities from humans and, over time, begin to replace them.

**AI role: **Here we describe the main capabilities of AI and its relationship to a human colleague by level. This is from a non-technical perspective. We’ll follow up with a deeper technological perspective on each level if there’s interest but for now we wanted the relationship to stand out.

Human role: This is the flipside to the AI role, again focusing on the relationship between human and AI and their relative responsibilities and capabilities.

Adoption: This is simply the date at which we expect mainstream adopters (broadly encompassing both early and late majority adopter categories)  to start applying AI at each level. Innovators and early adopters will be earlier still and the laggards will likely be later unless and until a crisis changes their trajectory.

We know that adoption rates are going to vary from industry to industry and from department to department. Even at the employee level, it’s highly unlikely that adoption will be a smooth process. Some individuals will readily embrace AI, although they are more likely to embrace the AI that frees them from the monotonous and boring aspects of their job than the AI that promises (or threatens!) to perform the more creative and/or strategic parts.

Others still, especially those who fear that their job will be completely replaced by AI, are likely to push back against the whole thing. Broadly speaking, though, we’re already seeing examples of both predictive and generative AI being applied across most industries and we know that more sophisticated and capable bots and agents are coming soon.

Autonomous work implications for business

We’ve identified three important implications of this AI evolution for business and we hope that leaders will recognize that they’re on the horizon and arriving soon, and start to plan accordingly:

• **Beyond human capabilities: **AI will not merely progress to being more productive compared to individual human full-time equivalents (FTEs) or being measured in manpower units (as we discussed in our previous article on AI, horses and humans). At levels 5 and 6, AI will demonstrate the ability to handle situations beyond the abilities of any number of humans. It will then be measured in machine power which will not be simply in terms of GPUs/CPUs or Transactions Per Second (TPS) but probably as some function of complexity, accuracy, and speed.