By now we’ve all heard about AI and machine learning. Many already know the huge benefits that come with implementing predictive maintenance and automated product inspection. But we still need to be careful upon entering, since we are still navigating a place with little rules and many unknowns, but with huge potential. Welcome to the wild west of Artificial Intelligence.
Forget about logic
In almost every case IT follows a specific order. If you fill in this, or do this, then this will happen, appear or move. AI works completely different. Forget the logic, AI revolves around patterns. Investigate this and see what happens. If you’ll find a pattern, you may have gold. But if not, you have to try again.
What are you trying to solve?
This trial and error way of working can be highly labour intensive and therefore expensive. In order to make the life of our data experts easier, it is really important for them to have the right specifications. What kind of problem are you trying to solve? What do you want to achieve? And, very importantly, is AI really the answer you need?
Just do something, ok?
If a data expert will start working with only your requirements it often goes wrong. This is like your boss asking you to write a story. Just a story. Of course you will be delighted to write him or her a story, but in order to do so you need the right information. What should be the subject? How long should it be. What style should it be? And so on. Without that it’s almost impossible to achieve the right result. The same counts for successfully implementing an AI solution for your factory or industry.
In order to increase your change of success, there are a few tips that may help.
- Ask yourself the right questions. Is AI really what you need? Are the potential benefits worth enough to risk implementing AI? Like in the real wild west there are no guaranties for success. At least not at the beginning.
- Work iterative. Start with your strategy and a concept, then, continue with creating a Proof of Concept to check if your plans are realistic (trial and error) and only then push the button and go for an MVP (Minimum Viable Product).
- To lower your risk even further the data team should work as closely as possible with the (key) users.
Close collaboration is key
To explain this last point: at LINKIT we use the slogan ‘building it together’. This is especially important when implementing an AI solution. Remember the part about just writing a story. With AI there are always numerous factors which have to be taken in account. And to make it even more difficult, they are different for each situation. Things like weather, lighting, machine specifications and even how people are using them will seriously effect the outcome of the underlying algorithm.
Getting all the information you need is very important as a data scientist or engineer. Therefore, close collaboration with machine operators, among others, is key for achieving the right result.
Do you want to know more about the ‘how’ of implementing AI and machine learning in an industrial environment? You can read all about it in our whitepaper ‘From unstructured data tot business value for manufacturing’. Or you can ask one of our experts for more information.