Under the abbreviation “K.I.” for artificial intelligence, countless applications, technologies and concepts – and also completely misconceptions – are summarized.

The concrete meaning, however, is actually quite simple: whenever a machine uses procedures modelled on those of our human brain, it is artificial intelligence.

Understanding Artificial Intelligence

When the average citizen hears the term, he usually thinks first of sophisticated robots that spectacularly destroy the Earth in bombastic action scenes in Hollywood movies. The only thing this image has in common with artificial intelligence is the technology used in the animation studios to produce the film.

Instead, the use of AI (or the more common English term “AI” for “artificial intelligence”) says nothing about the shape or application of a device. Even an inconspicuous app on our phone can meet the conditions of artificial intelligence. Because it only counts how a program works: the use of thought patterns, which are self-evident for us humans, such as logical thinking, drawing conclusions or creative problem solving, defines an AI.

One can imagine that it is not easy for a machine to learn logic. A computer, for example, masters only one activity from a technical point of view: it can perform calculations. If you give him input in the form of a calculation task, it returns the output in the form of a result. All the other modern possibilities to which we have become so accustomed, not least through digital transformation, are based on this simple basic function.

However, it’s a long way from performing simple calculations to intelligently detecting, understanding, and solving complex problems. Simple, non-intelligent computer systems often used solutions such as brute forcing to solve problems through billions of computing operations by calculating all possibilities until a successful one could be found. Instead, an artificial intelligence-based system is trying to find the right solution without trying all the wrong ways.

Due to the progressive digitalization of our society, the increasing networking and the favorable availability of high-performance computers, the way has already been paved for the use of artificial intelligence in all areas of life.

Definition of digitization

Digitalization is, quite soberly speaking, simply the transfer of formerly analogue processes to digital ones. Even if we are currently increasingly encountering these and similar terms, this is a very old and simple process, because almost every form of digitization is rewarded with efficiency increases, cost reductions and new, previously unknown possibilities. No wonder we humans have always been very interested in her.

Due to the accelerating technical progress and the mutual support (new technologies enable new technologies …) digitalization has gained so much speed in recent years that it has now penetrated into all areas of our lives and is indispensable from there. This digital transformation is a technological, socio-cultural, economic and intellectual process that brings with it gigantic upheavals.

For companies in particular, the digital transformation creates unprecedented opportunities – but it also lurks with considerable dangers, especially if it is ignored.

Demarcation of AI to other systems

Which types of applications are considered and which are not artificial intelligence is often disputed in practice. Also, tasks that were considered groundbreaking in the past, but have meanwhile become the absolute standard (for example, automatic image recognition), may lose their “AI status”.

Similarly, attempts to make a distinction based on the solution are failed: although it is true that an AI does not find the desired answer by pure computing power and try out all possibilities; However, since it is also (and in some cases even more dependent) on extremely powerful computer systems, the boundaries quickly blur here as well.

Due to the blurring that is always present in the determination of AI systems, misunderstandings can quickly occur. In practice, therefore, naming the concrete method (“We use a neural network for ….”) has become naturalized over the use of the AI term (“We use an AI for…”). Nevertheless, a lot of patience is often required when talking about artificial intelligence in a company.

Applications

Even if it is a very strained phrase, it fits perfectly here: the possibilities are unlimited. Due to the enormous potential of artificial intelligence in our already extremely networked and digitized society, there are areas of application everywhere. In addition, advances in AI in turn lead to an acceleration of development, creating an exponential growth effect.

In our private environment, the effects are already noticeable – although not always directly visible. Artificial intelligence allows Alexa to understand our instructions and makes Google’s translation app a little better every day.

Due to the extreme adaptability of artificial intelligence, it would not be expedient to list potential benefits for individual industries. Instead, the following are concrete applications and potentials that can be realized by AI – regardless of the business area in almost every company. Open

Analysis

Data is the most important asset of a modern digital enterprise – its existence and quality determine the ability to act customer-centric and compete in the long run.

The analysis of such data is one of the most interesting uses of artificial intelligence. While extremely valuable insights are already being gained by humans in the areas of business intelligence and data science, AI’s use provides new, hitherto unknown insights. The non-human nature of computer systems can be used to identify connections that remain hidden from the human viewer and his equally human patterns of thought.

For example, the analyst of an online retailer would immediately realize that the sharp increase in the number of TVs sold is related to an upcoming World Cup – and that, once it has ended, you have to adjust to increased returns. As a result, the online retailer would probably adjust its business rules, such as increasing shipping costs for electrical appliances or rejecting new customers who only order a TV. However, the part of the “honest” customers who are looking for a new device and want to keep it permanently would be lost; at the same time, however, you save yourself returns-related failures. A necessary evil.

Thanks to the analysis of historical data, however, further insights could be gained by the AI use. For example, the screen diagonal and the sales price may be used to determine that certain devices or device classes are less likely to be returned. Buyers of such devices could then be encouraged by an installment payment offer to keep the new TV after the World Cup. This allows the company to make the right decisions, to circumvent the risks (in this case, returns), and ultimately to generate new revenue, even in a difficult environment. A direct added value that could be realized through AnI-based data analysis.

AI technology excels in the analysis of unstructured data, which according to common estimates account for 80-90% of all information available in a company: this treasure trove of data is normally inaccessible to standard processing methods, but can be examined by artificial intelligence. The potentials and insights gained are often extremely far-reaching.

Structure unstructured data

As mentioned in the first point, by far the largest amount of information within a company is available in unstructured form. This data is usually very text-heavy and not indexed, so it is not known where a document on a particular topic can be found.

It does not take much imagination to imagine that the use of this information would only be scoured with gigantic human effort. Even “classic” computer programs have no chance here: they could search all files for special keywords and then group them accordingly. But that would require you to already know all the relevant keywords and topics – and that’s almost impossible, at least in larger companies. Other options might be to create a Wordcloud, but it could only tell you the most common words… how to turn and turn it: without AI, you are denied access to this information.

By using artificial intelligence, on the other hand, it is possible for the first time not only to examine files for certain characteristics, but also to have the system read and understand them – completely open to results. Similar to how Google today, in response to our search query, does not just provide a list of websites that have the same words, but tries to understand and answer our question (as “Plan B”, if the question is not understood, Google is again searching for words – but that’s a different story).

Since an AI then knows and understands our unstructured data, it can make different sortings for at the touch of a button or classify it by topic. Also information that does not contain corresponding keywords, but is still related to our searched content, can be recognized and assigned in this way.

If all the intellectual capital of a company is to be used, there is therefore no way around the use of artificial intelligence.

Recognition and mapping

One of the most well-known features of artificial intelligence is image recognition – a system that recognizes, for example, human faces and their emotions, as well as various objects in images or videos. The concept can be described, very generalized, as “assigning meaning to a bunch of pixels”.

However, the application is not limited to optical media; the fields of application in companies are enormous.

For example, an AI can check legal department contracts for errors and loopholes, issue alerts, or suggest the best content category, the most appropriate emojis, and the ideal release time to generate maximum reach.

The most energy-saving heating and lighting plan can be created using sensors in the company building and, thanks to the automatic identification of the request, the caller can be transferred directly to the right contact person in the call centre. And because this is not enough, artificial intelligence also recognizes the customer’s mood by means of the voice and proposes suitable texts to our employee in real time, which have a de-escalating or sales-promoting effect (and whose effectiveness has of course also been determined by AI-supported).

In Japan, insurance agencies have successfully tested the use of AI in the analysis of accident images to identify where there is real damage and where it is an attempt at fraud. The CEO now uses Siri, Alexa and Co. to dictate his email, and the management’s assistance lets a AI decode the illegible handwriting on a notepad – something that the computer already controls more than twice as well as any human being.

The list could be continued endlessly and extended to every division of the company. Wherever something needs to be recognized and categorized, AIs have the potential to generate high added value. In addition, if there is a correspondingly large amount of data from which artificial intelligence can learn, the results improve drastically.

Disadvantages and dangers

AI use stands like hardly any other field for the enormous possibilities and effects of digitization. But again, not all gold is that shines.

Significant disadvantages arise primarily from the complexity and relatively little experience that has been gained so far with the technology, which is still quite young. The optimism expressed by enthusiasts also has the potential to cause disillusionment or disappointment.

Complexity and cost

Using an AI in a predetermined frame is extremely easy today. If you have a concrete, manageable use case (“We want to scan our old files and convert them into text files”), you will find providers or software that can implement your request out-of-the-box. If, on the other hand, you want to use artificial intelligence intensively and according to your own parameters for your needs, you have to build up the appropriate expertise in the company – and that will be expensive.

AI specialists are among the most sought-after people today. Of course, this is also reflected in the price: a graduate of a well-known university, without professional experience, will not even give your job offer an answer, unless at least €200,000 is stated as the basis for negotiations.

This presents small and medium-sized enterprises with seemingly insoluble tasks. The training and further education of our own employees is therefore a popular alternative. Although this is not exactly cheap, it can be booked at a much lower cost. Since suitable candidates, in addition to software development, often come from the field of data science, a considerable group of competent specialists has been formed here in recent years, who benefit from a very open exchange of know-how. The training offers are also very productive at the moment.

However, building up AI competence remains a very expensive and time-consuming undertaking.

Excessive expectations

The tech world is literally overflowing with daily news about the latest advances in artificial intelligence. In the concrete application, on the other hand, disillusionment quickly occurs: although the potential is really unlimited, in reality it often fails with usefulness.

The main reasons for this are usually a lack of data base (non-existent, qualitatively sufficient or inaccessible for bureaucratic reasons), unsuitable corporate structures and processes or simply a lack of know-how.

Also, the fact that the technology has been used for a completely inappropriate task is a common conclusion to be heard after failed projects.

Last but not least, it should not be forgotten how quickly artificial intelligence can fail in the defense of individual employees – especially if they take leadership positions and consider digitization to be a temporary phenomenon that needs to be sat out.

Conclusion

As the technology that will influence and already influence our lives like no other, the importance of artificial intelligence cannot be overestimated. However, access remains difficult for many companies due to sometimes high complexity – although the actual difficulties often remain well below what is spread by hearsay.