Artificial intelligence, AI for short, is the current buzzword of the CPM industry, even before the block chain.
Last week, the topic was high on my agenda: an article on the topic for the current is-report had to be written, a customer workshop had to be held from which a new AI system will be developed, the AI system of an existing customer had to be prepared for the new run and, as a preliminary highlight, a theme evening on "Robotics and Artificial Intelligence". Reason enough to write down some personal thoughts about the current state of affairs - not only, but also with regard to controlling and CPM systems - in a blog post.
Strong AI - in the sense that an artificial intelligence can really think, is self-conscious and could catch up with a human intelligence on an equal footing, is still not in sight. Whether this will ever happen, and whether it is even desirable, I doubt.
However, weak AI do exist - basically they are algorithms that know how to handle large amounts of data on the basis of rules and can optimize themselves within limits. And we encounter them in our daily life - be it Amazon's famous recommendation system, Siri, Alexa & Co, the driving assistant in the car or chat systems, where the AI tries to interact with users / customers to such an extent that the need is satisfied in the simplest case (how will the weather be tomorrow) or the request can be forwarded to a qualified person. This can be fun or even annoying.
Robots, as sci-fi fans know them from Star Wars or Isaac Asimov's works, are not in sight. Prepper, the humanoid robot is a weak AI in a humanoid-shaped machine and uses the childlike scheme to win sympathy. Why shouldn't it?
Even if the actually available AI is only a weak one - it still offers a wide range of potential uses, including business applications. With realistic expectations, a lot can already be achieved today. However, the marketing materials usually do not show that a good AI is preceded by the development of the system. In the paths in which AI can "think", it is as fast as an arrow and powerful, but the paths and parameters must be defined. A current project example can be found in the article of is-report described.
It is important to us to promise only as much as is feasible in the project. The challenge is to simplify this process of parameterization of AI models and to standardize it as far as possible. Recurring use cases can lead to industry solutions - further modules for the CoPlanner framework. In cooperation with AI experts and the customers, especially from the (controlling) departments, we translate technical requirements into technical solutions.
Will a competitor overtake us on the left or right side of the track? In the marketing promise, maybe - I know AI "systems" that read great as product information and for which marketing is way ahead of implementation. We'd rather come onto the market a little later with such modules - but they will work. And in the meantime implement individual solutions with and for our customers, which are tailored to the respective situation.