Right now, our training on digitalisation is going on and one of the things we’re talking about is automation. Automation is a natural part of digitalisation, that’s for sure.
Digitalisation is to some extent a buzzword. After all, we have been working on introducing “IT” into our business for a long time. However, it has been more about acquiring a large number of business systems. These are financial systems, customer care systems, planning systems, and others.
Lack of drive
These systems have very little “drive” in them where we are supported in working through an entire process. They are more both strongly linked to the respective department and also intended for manual management within their domain. They include good management of different types of data, and also smart functions around this data. But as I said, no drive forward in our process and interconnection with other necessary systems along the process flow.
I think what really sets us apart today, versus in the past, is that we are looking more at how we can on the one hand achieve control throughout the process (from start to finish) and also how we can achieve automation in those process flows.
Opposition to the transmission of the decision
If we look at the latter, there is a built-in resistance to letting the computer make decisions for us. Especially in the parts that we feel are our core business. We don’t really trust that good decisions will be made if we humans don’t make them. In my experience, however, in most cases it is the other way around.
It is important that the automatic decisions taken in public life do not deprive us of the transparency that a democratic society stands for. It must not be the case that the automatic decisions become non-transparent, where we do not understand what is programmed into it or what algorithm is used.
In my experience, decisions that are automatic are often more understandable and legally certain than those made by people. I have met businesses where it is known among their customers which administrator to talk to in order to get what they want. This is because one officer is identified as “easier to deal with” than the other.
Crystal clear is a must
When decisions are digitised, one is forced to specify how the decision is to be made and what the basis is. If it is to be automated, it must be crystal clear what the basis for the decision is.
Many times when I sit with these issues, I see that it is difficult from the business side to explain what is the basis for decisions. It then differs between different people’s views where there is disagreement.
The strength of us humans is to be able to make decisions on intuition, but if it differs between us, which it does, then the decision will also vary.
Digital decisions are always the same, but can lack human insight. A combination is good to aim for in cases where we leave the simple and obvious decisions to the computer and the more complicated ones where human judgement is needed to us humans.
As AI becomes more widely used in our decision-making engines, we will see that even decisions of a more difficult nature can be made automatically.
Prerequisites for success are…
A prerequisite for automating decisions is to have control over the value-creating flows. Because it is only when you see the flow that you can also clearly see the evidence that is gradually being built up for the decision itself. Then you can review how the decision itself is actually made today and also how you can automate it.
Another important prerequisite is that business people do not leave it to the “IT people” to develop the automated decisions themselves. Unfortunately, I meet businesses where they don’t understand how decisions are made themselves, but say it’s something the programmer can answer.
It is, as I see it, not the programmer’s fault, but that the business has left it all to the programmer to solve.
That’s not the way to have it. The business must have 100% control over how it specifies, develops and verifies the automated decisions.
Until next time; think about the decisions you make in your business, i.e. identify them. Then think further about whether and which of these can be automated.
Greetings,
Matts
