The Six Levels of Autonomous Organisations

Have you ever stopped to think what cars might look like once a human is no longer required to control them?  The design constraints imposed by human command and even combustion engines have dominated car design for over a century.  With all that gone, we’re once again free to rethink how we use, and ultimately design, the humble automobile.

But what about our businesses? To date, organisations have been designed specifically around the way we make and communicate decisions.  However, with the cost of global communication essentially falling to zero, traditional structures are already starting to fall out of favour. 

Limits on these structures now lie very firmly with our abilities as humans.  Like it or not, no arrangement of the deckchairs will overcome those constraints. But a machine can. 

What we’re talking about here is business decisions made not in days or weeks, but in milli, or even nano seconds.  You can’t have that level of response with a human in the middle. If humans are removed from the control of our organisations, then well, what then?

The 6 levels of Autonomy

At one level self-driving cars are paving the way for a broader ecosystem of autonomous things.  The scene is being set for a whole range of socioeconomical issues; Legal ownership, liability, regulation, ethical concerns, impacts to employment, the list goes on. 

Can self-driving cars then open a window to what the future of business might look like?  If they do, then the levels of autonomy might be a good place to start.

When it comes to self-driving cars, the levels of autonomy has become a fairly standard set of definitions.  The 6 levels (from 0 to 5) allow us to establish a frame of reference for autonomous systems.  Beyond just a set of ambitions, they help us to understand the impacts autonomy will bring.  Hopefully allowing us to plan appropriately for what lays ahead.

Understanding the impacts of autonomous organisations is valuable for the same reasons.  Perhaps then those same basic levels will make sense here.  In which case, they might look something like this:

• Level 0: Every action and decision in the business is performed by a human.  I’m sure we all know at least one business that still seems like this!
• Level 1: Business are controlled and operated by humans, but with certain functions performed with the aid of machines.  An automatic coffee machine for example
• Level 2: Business still controlled and operated by humans.  Under a very limited set of circumstances, some functions of the business are being performed automatically by machines, such as in a production line.
• Level 3: The business is guided by humans, but most, if not all operational activities have been automated.  Some business-critical decisions are now being made by machines under certain conditions, but still with a high level of human oversight. 
• Level 4: Full autonomy of the business exists, but only in certain, restricted problem spaces.
• Level 5: Machines have completely replaced the need for humans across any business domain or function.

A view from reality

I would suggest that on average, most businesses today are sitting somewhere between level 1 and 2.  We are actively using machines as tools, but very rarely letting them make autonomous decisions on our behalf.  Even then, those decisions tend to be constrained to a specific action rather than a broader, more critical business function.

Level 5 autonomy, however, is probably a long time away. Machines would have to reach the full AI capability of humans for this to happen.  Even if a fully autonomous, human free organisation became a reality, would society allow it?  Somehow, I doubt it.  Though visions of sky net do spring to mind.

What’s more likely I think is level 3 autonomy, or a semi-autonomous organisation. A human, or team of humans at the centre of a mesh of automated functions.  Humans playing the role of governor.  Setting the boundaries of operation and intervening if it all goes pear shaped, but little else.

You could argue that some businesses are heading at least towards level 3.  Algorithmic trading might be one example that applies here.  I’m no expert in this area but it would seem theoretically possible to create a nearly human free organisation today.  An organisation that exists solely to trade certain kinds of financial assets on an open market.

The Transition

Current leading businesses are already transitioning towards structures that emphasise specific outcomes rather than functional silos.  In a zero cost, zero latency model of an organisation, this is likely to make more and more sense. 

As impersonal as it may seem, no requirement to hire, fire, or reskill an autonomous entity. Capability units are simply traded in an out of the organisation based on need and the value they provide. Not unlike how a baseball team might be constructed.

Let’s just assume that is the end game here for the moment, we will still face many of the social, emotional and functional challenges we face today.  The need to establish chains of trust, integration, renumeration, and delivery certainly won’t change.  Rather than just a connection between humans, or machines, we must now consider the hybrid.  How could a machine for instance, trust a human?

From a governance level, we need the right processes in place to ensure machines are making the right choice, or to overrule them when they are not.  As a society, we need to be comfortable with questions of things like liability.  Who will be ultimately responsible for the consequences of an autonomous business decision?  Are there changes to corporate law required to facilitate all this?

Realistically, we’ve been battling these issues since the first industrial revolution.  Unlike the industrial revolution however, we’re now relying more and more on machines to make decisions on our behalf.  Autonomous cars have become an important test case on how we as a society, should react to this.

The pundits have Level 5 autonomy in cars in or around 2030.  Assuming businesses track in a similar way then the time to think about this is now.  Once the machines are making these decisions for us, then we may never get another word in.