One of the ultimate goals of artificial intelligence is the ability for machines to operate on their own, with little or any human interaction. This idea of autonomous systems makes up one of the seven patterns of AI that represents the common ways that organizations are applying AI. While some of the patterns are focused on predictive analytics or conversational patterns, or systems that can recognize things in the world around us, those patterns still involve human interaction. After all, we need humans to be involved in conversational or recognition systems. However, the autonomous pattern is much more complicated as we’re asking a machine to do something in the real world without a human in the loop. These sorts of systems are harder to implement and generally take longer to show ROI. 


The Autonomous Pattern explained

Autonomous systems are defined as systems that are able to accomplish a task, achieve a goal, or interact with its surroundings with minimal to no human involvement. It is also essential that these systems be able to predict, plan, and be aware of the world around them. This is applied both to physical hardware autonomous systems as well as software autonomous systems (software “bots”). 

Examples of this pattern include autonomous vehicles and machines, and autonomous bots of all sorts. We also see autonomous systems in the form of autonomous documentation and autonomous knowledge generation as well as autonomous processes and cognitive automation. Autonomous systems have the ability to create legal and medical documentation, invoices, and automatically log data. It can also help businesses automatically route tickets or workflows. With its ability to help with inventory forecasting, improve shipping times, and tracking, it’s easy to see why it’s talked about in a positive light around the area of logistics. 

 Collaborative bots, or cobots, are also a form of an autonomous system. In case you’re unfamiliar with this concept, cobots are meant to operate in conjunction with, and in close proximity to humans to perform their tasks. This is in contrast to industrial robots that are caged off and physically separated from humans. Even though cobots can operate in an augmented intelligence role, they’re meant to operate independently of humans even though they’re in close proximity. 

The primary objective of the autonomous systems pattern is to minimize or eliminate human labor. When removing a human from the equation, you need to make sure that the autonomous system is as close to human level performance as possible. As such, you can see that’s why this is one of the harder patterns to implement. Naturally, as autonomous systems aim to minimize human labor, they need to be reliable, consistent, and of a very high standard. 

Levels of autonomy

When many people think of examples in the autonomous pattern they instantly think of autonomous vehicles. After all, to have a vehicle that fully drives a human from point A to point B safely without the human ever having to get involved or take over driving, would be incredible and is something we can all imagine. However, as we move from vehicles that are fully operated by humans to fully autonomous it’s not an all or nothing approach. 

The Society for Automotive Engineers has defined six levels of autonomy from Level 0 to Level 5. At one end of the spectrum, level 0 autonomous vehicles require full human control of all operation and decision making. At the other end of the spectrum, level 5 fully autonomous vehicles will bring dramatic changes to most industries and will have major societal and economic impacts. Levels of autonomy are necessary due to the high-risk nature of real-life autonomous applications. There is little room for error so further precaution must be taken with each level. Incorporating autonomous systems into a project or real-life situation is a complex and difficult task and autonomous applications need to be as close to perfect as possible.

When people think of the autonomous pattern they also think of robots such as Rosie from the Jetsons, C3PO or R2D2 from Star Wars, or other robots coming from Hollywood. These robots are able to perform many tasks that humans can such as talking, cooking, picking up objects, avoiding obstacles in front of them, and an array of other things that humans do. However, in reality, many robots have varying levels of intelligence, if any. Industrial robots, for example, are simply machines that are programmed to perform a repetitive task. The vast majority of robotic systems contain no machine learning system or intelligence so to speak. These robots are programmed to complete repetitive and laborious tasks with the aim of decreasing human labor while performing various tasks more efficiently. In essence, this is automation, allowing for a process or procedure to be performed with minimal human assistance, but containing no intelligent features. 

The oft-overlooked fact is that automation does not correlate to intelligence. Just because a system can repeat something that a human instructed it to do does not mean it has the capacity to learn anything. These machines are simply programmed to perform the same task over and over again. The misunderstanding of automation as being something more than just human instructions and prooramming could be due to the general public’s tendency to envision the version of robots that the media portrays upon the mention of robotics.

This leads to the discussion of software automation as well as hardware. The popularity of Robotic Process Automation (RPA) is real. People want to be able to automate software-based processes that ordinarily requires a lot of human activity. Despite RPA’s lack of intelligent processes, the tasks they perform are vital in helping companies to ameliorate their cognitive intelligence systems. Moreover, these machine processes significantly lower labor costs and can show immediate ROI despite their lack of AI. Indeed, many consider RPA a gateway to AI. 

RPA doesn’t have to remain strictly AI-free though. Just as AI can be added to industrial robots, it can also be added to RPA. An asset of this collaboration would be the ability to create intelligent workflow automation which could automatically detect changes in software or systems and work bottlenecks. Along the same vein runs the possibility of developing autonomous business processes that would allow for optimization and analysis of business processes without the need for humans. Over the past several months, a few of the large RPA vendors are indeed looking to incorporate AI into RPA.

While the autonomous patterns system of AI is arguably the hardest to implement, when done correctly, it can have incredible impact.  It is recommended that much thought be put into any step that might involve autonomy. With the many possibilities, the future of the autonomous pattern is a bright one.