Will replace AI data science

Artificial intelligence: why it cannot (yet) replace common sense

Dr. Jürgen Cito from the Computer Science and Artificial Intelligence Laboratory (CSAIL) of the renowned Massachusetts Institute of Technology (MIT) was at the invitation of WKS and AUSSENWIRTSCHAFT AUSTRIA (AWO) of the WKÖ in Salzburg and gave an insight into his projects with IBM, Facebook and Boeing, among others . Around 20 entrepreneurs, startups and researchers from the Information Technology & System Management course at the Salzburg University of Applied Sciences exchanged information on the state of technology and current challenges.

The MIT researcher first cleared up some myths. “Artificial intelligence is still a statistical approach. It cannot currently replace common sense or imitate the human brain. Creative decisions still have to be made by people, ”says the native Austrian. He prefers to speak of augmented, i.e. "supplementary intelligence" that supports technicians in their tasks and, in the best case, takes on annoying tasks. Such as data preparation: Data scientists would spend 80 percent of the time on the unpleasant task of cleaning up chaotic databases before they are even suitable as a database.

Package insert for data

That artificial intelligence is objective is also a myth, according to Cito. If a subjective selection has already been made in the underlying data sets, this is retained. One possible solution would be “package inserts” for data that provide information about how a data set can or should be used. But what if such data sets are only available to a limited extent, as is usually the case in small and medium-sized enterprises (SMEs)? Jürgen Cito emphasizes that you need a huge amount of your own data is another myth. Research is being carried out not only at MIT, but also at the Salzburg University of Applied Sciences.

Small Data: The Challenge for SMEs

"We are concerned with possible applications specifically for Austrian SMEs," says Stefan Wegenkittl, scientific director of the Applied Image and Signal Processing course and head of the Applied Mathematics and Data Mining course in the Information Technology & System Management course. On the one hand, his team is working on supplementing existing, small data sets with artificially generated data and thus creating a larger basis that is required for the development of artificial intelligence. On the other hand, models based on large data sets from the big players can be transferred and adapted for special applications with little data. "This is how we make artificial intelligence techniques accessible to SMEs."

Much educational work needed

During the subsequent exchange, the invited business representatives agreed that there is still a great need for clarification in Salzburg companies about what artificial intelligence can and cannot do. It was discussed how internationally established research results could be integrated into the domestic economy. Thomas Heistracher, head of research at the Information Technology & System Management (ITS) course, emphasized that you don't have to wander far into the distance: “In terms of content, many of our subject areas overlap with those at MIT. A sign that we are researching on the cutting edge at the University of Applied Sciences. "

Data Science at the Salzburg University of Applied Sciences

The ITS course at the Salzburg University of Applied Sciences offers prospective data scientists with their own specialization in Data Science & Analytics as part of the Master’s course optimal training. In addition to the mathematical and statistical core competencies, graduates have in-depth knowledge in the areas of machine learning, data mining and deep learning.

it's MAGAZINE

You can find out more about artificial intelligence in the next issue (September 2019) of the degree program's magazine. Order the print version free of charge from [email protected] or read the online version at www.its.fh-salzburg.ac.at/its-magazin.