THE BLOG

Overcoming The Paradox Of Automation

15/08/2017 16:09 BST | Updated 15/08/2017 16:09 BST

In a nutshell, the paradox of automation is the idea that the more advanced automated systems are, the more critical any human involvement becomes.

The Air France Flight 447 tragedy has become synonymous with the concept. Perhaps because I'm a nervous flyer, or maybe as it was the first time the paradox was imprinted upon the public psyche, the story really resonated with me. Nothing else has so tragically demonstrated the narrative that in the rare instances an automated system should fail, it will invariably be an intense and extreme situation. As such, the requisite manual skill must be excessive - in this instance, being able to prevent a faltering plane from dropping out the sky. Unfortunately, on May 31, 2009, the pilots were unable to do so.

And yet, they could have easily gone their entire careers without ever needing to.

Cases like this are of course extraordinarily rare, and follow-up research made me a little more comfortable; every other pilot who was challenged in simulated recreated environments handled the situation with aplomb. This just happened to be the perfect storm. Literally.

So, as you might imagine, I soon after turned my attention to how such instances become avoidable. Yes, the software should be unbreakable, but we cannot always account for those extreme and unusual circumstances.

The Implications for Modern Businesses

Contained within the paradox is the necessity for manual expertise. We can indeed take this one step further, as not only must the operator be extremely adept in their profession, they must be able to retain their expertise after lengthy periods of inactivity, beyond overseeing and monitoring automation systems.

Clearly this creates risks, but it also generates opportunity. Automation is a great way to remove the more tedious and menial aspects of a job and as such we can entrust the computers to run basic procedures. It is typically faster and more reliable than human input, which gifts us time - time we can spend constructively. Ultimately it is crucial that we use this to prepare ourselves for those unprecedented situations. Because if we don't, catastrophe awaits.

Of course, the easy (and often cheap) temptation is to let complacency seep in and allow the machines to take care of everything. Or worse yet, use automation to mask incompetency. Unfortunately, in some sectors, this does happen. In the ruthlessly competitive world of business, why would a company pay a large amount of money to have an expert sit around doing nothing other than watch a machine. And if they did, what industry-experts would get job satisfaction from sitting around doing nothing all day? It is almost a paradox within the paradox.

Finding a solution to this can seem challenging, but putting blind faith in an unthinking machine is far more dangerous. One avenue that can be explored is to continually engage the operators of the machinery. Perhaps the machine should be the failsafe. This way we continue to complete tasks as if there are no machines, but when a human makes a mistake, the machines corrects. In this scenario, the automation should be augmenting human processes, rather than the reverse.

However, while this would theoretically help, there is a valid counterargument that in some instances, it can create an illusion - because eventually we will learn that we can get away with anything if the machines keep correcting our errors. This can be mitigated by ensuring operators address small issues - not allowing the automation to proceed unless a manual action is taken. While this would have a minor impact on business efficiency, in the long run it can certainly help.

A further option is to increase the amount of testing. Allow the experts within an organization to innovate and explore the extreme situations automation is not built for, subsequently redesigning the machines to overcome these issues. In a more physical situation (such as an aircraft), the operators themselves can also be continually tested. They must be competent taking over in extreme situations. Obviously, this has its limitations, as the unprecedented circumstances in which automation might break will typical be more or less unique. But it can mentally engage and prepare them as best as possible.

The truth is, no matter how automation is dressed up, the more efficient it is, human involvement becomes more important not less. The challenge for the modern business is to ensure they have that expert human input as and when they need it. The methodology might vary from organization to organization and expert to expert, but it can be the difference between success and failure in more ways than one.