AI against Artificial Intelligence — Buzzword against Reality
an article by Tim Oberlies and Marco Sälzer
Artificial intelligence — hardly any other term is so much Buzzword as AI. We at brain!act are surprised in how many different magazines you can read about AI. Let’s make an experiment: Look for a random article on the subject of digitization. Is AI mentioned in this article? If so: We said it. If not: You probably found what the data scientist would call an outlier.
This phenomenon is supported by the fact that AI rarely occurs on its own. Usually, the term is used in connection with others such as data science, computational intelligence or big data. The question that arises unavoidably: How much artificial intelligence is actually behind the term AI?
And this is where it gets difficult: First of all, a definition of artificial intelligence is needed. However, one quickly finds out that there is no worthy and/or widely accepted definition of artificial intelligence. Let’s make it easy for ourselves and decide that we see it as a collective term for various techniques and algorithms that generate knowledge, make forecasts or make simple decisions on the basis of relatively large amounts of data.
Obviously, this definition ignores claims such as autonomy, consciousness, and free action or even morality and ethics. Of course, this spoils the term of some enthusiasm potential, but it is currently the most consistent and realistic definition. However, one quickly realizes that even from such a realistic viewpoint there is still much more called AI than artificial intelligence actually means. Already exaggerated expectations are missed even further.
Simple statistical analysis methods, applied to already existing data sets, deliver fast and in the first place good results, but quickly reach their limits if applied to changing applications and dynamic data. Automation of existing processes is mistakenly called AI, which on closer inspection leads to disappointment and rejection of the actual potential of artificial intelligence.
But is it only a problem of definition or understanding that so many AI projects are wrongly evaluated, give away their potential or even fail? No.
It becomes clear if we think about the interfaces at which a typical AI project has to dock on: It needs
- expert and domain knowledge
- understanding of the existing or required data and their storage structures
- knowledge about data preparation and preprocessing
- best-practice skills regarding algorithms from the field of machine learning
- sound knowledge from the fields of statistics, probability theory, and algorithms
A team that does not specialize in data science or computational intelligence will hardly be able to meet these requirements. It is, therefore, a competence problem, too.
And this is where brain!act comes in: We know that it is the right way to look at possible use cases in the company first before a large-scale AI project is launched. In a workshop, we jointly highlight possible starting points that fit the respective company. As a result, brain!act brings science, technology, and business together in such a way that added value is created for all participants without generating and then missing exaggerated expectations.
Areas of application can extend over the entire value chain of a company. Whether in research & development, production, sales, logistics or service, companies can use various AI technologies. An example situation from goods receiving: The materials are often manually booked into the merchandise management system in order to start the intralogistics workflow. Image recognition (a typical use case of artificial intelligence) could fundamentally improve this situation. The truck would have to open the tarpaulin and the material would be booked in automatically. As a result, the company could save time and labor.
In addition to the interest in using the technology to an appropriate extent, there is also a cultural factor that we actively tackle with this approach. Through our approach, we improve the information situation in the company and prove the understanding of the technology behind AI through an integrated implementation project. This creates trust and a tangible starting point to establish AI in the company as a technological capability.
Nevertheless, we do not want to stop at this point. As already indicated, we are faced with a challenge: To make the technologically complex and interdisciplinary topic of AI accessible. The vision of brain!act is to encapsulate best-practice solutions technologically and methodically in a way that new models can be trained or related to new problems with little effort. This demands, however, that we collect knowledge about needs and prerequisites in projects with AI inexperienced companies.
If you are interested in becoming one of these brain!act pilot companies, contact us.
