Das ARIC hat einen neuen Vorstand. Neuer Teil des ARIC-Vorstands ist Michael Koch, Director Data Analytics & Artificial Intelligence bei Lufthansa Industry Solutions. In Teil 1 des ARIC-Interviews spricht er darüber, wie er zu KI gekommen ist und, welche Entwicklungen er im KI-Bereich beobachtet.
How did you become interested in AI?
Michael Koch: I was already interested in data and large database systems when I was studying computer science. In my first few years of work, I then set up central data hubs for various companies in the industry. The main focus was on evaluating history and generating decisions from it. When I started at Lufthansa, it was increasingly about making valid predictions, such as how large the cargo volume per aircraft would be or how many meals would have to be planned on board. And this is where AI increasingly came into play.
What other exciting AI applications are there?
There are many possible applications for AI in companies, especially in light of the current pandemic. If, for example, computer vision is used to comply with distancing rules, image counting can be used to determine How many people are on site or how long the waiting times will be. This is particularly interesting in the airport sector, but also for tourist hot spots such as St. Peter Ording. Another example is speech processing, which can be used for the rapid automated evaluation of customer feedback. AIs can process this data very reliably in order to address the right contact person and shorten response times.
Tourist census by AI in St. Peter Ording
In the tourist resort of St. Peter Ording, LHIND is working with project partners on a digital visitor guidance project. The AI evaluates data from anonymized Wi-Fi tracking, anonymized image recognition and laser technology and measures the density of people and occupancy. The results are then displayed as a traffic light system – for example on visitors’ smartphones, who can then find out how busy the beaches are.
How has the AI sector developed in recent years?
In general, the rapid development of artificial intelligence is due in particular to methodological advances and the greater availability of mass data and computing power. The volumes of data that can now be collected, stored and processed are what make machine learning processes and modern AI applications possible in the first place. In a data-driven company, decision-makers can rely on data instead of gut instinct.
Trust in predictive systems has also grown and is now commonplace in larger companies. AI is now also reaching SMEs, as pre-trained solutions as a cloud service have brought business-generating AI within the reach of medium-sized companies. However, we are still nowhere near where we could be. Although companies often experiment with AI, they have difficulties integrating the technology into their standard processes. It is therefore important to think carefully about which AI solutions are economically and strategically feasible for a company.
Suppose an AI sceptic were to say: I don’t want everything to work via AI, it’s scary for me. What would you say back?
Don’t be afraid of AI, because it is mainly used to solve repetitive tasks. And these are tasks that most of us don’t want to do. For example, answering the same questions in emails for 1-2 hours a day or counting people, which can be done much more easily and quickly with an AI. AI applications support each and every one of us in our everyday tasks.
On the other hand, they can do nothing else: they are not hyper-intelligent. They can only perform routine tasks and are no better than humans. In the negative case, they even reproduce our mistakes and prejudices because they are trained with human data. This is why it is important to insist on data quality, because – whether intentionally or unintentionally – the respective analysis results can be distorted or manipulated by an arbitrary, inadequate or even discriminatory selection of input data.
So one opportunity for AI is to automate monotonous work. Do you think there is other potential?
By using artificial intelligence (AI), companies can achieve greater efficiency, make more informed decisions and develop completely new business models.
There are now business models that are based entirely on data. You can see this in the insurance industry, for example, where anyone can now take out a car insurance policy online. A lot of this is based on AI, which makes the right suggestions and connections. That didn’t exist 5-10 years ago.
In Teil 2 des Interviews spricht Michael Koch über den Mittelstand und KI und den Stand der Automatisierung in Deutschland.
