Artificial intelligence put to the test
- Digital Services
- Created by Julia Guthan, TÜV AUSTRIA; JKU
Artificial intelligence is one of the fastest growing technologies and is being used in more and more applications in the form of machine learning algorithms. No matter whether image recognition, recommendation systems, chat bots, diagnostics or forecasts – machine learning models trained with large amounts of data can already be found in a wide range of everyday applications, giving them “intelligence”.
Can I trust you?
It is precisely this appearance of intelligence, however, that is unsettling and raises several questions: Is AI trustworthy and safe? Does the use of AI bring the expected value added? Is the data used in the training correct? Is the large amount of sensitive data handled carefully? TÜV AUSTRIA is working together with the Machine Learning at the Johannes Kepler University on corresponding certification methods to help manufacturers in the development of safe, reliable and high-quality machine learning models and to provide users with a seal of quality for trustworthy AI systems.
The Director of the Institute, Prof. Sepp Hochreiter, believes that AI will set a precedent for the future: “Machine Learning is currently the most important enabling technology and will have a massive effect on our technical environment and our entire lives in the long term. This makes it more important to strengthen consumer trust in this technology through the certification of machine learning applications. Which is why we are delighted to be involved in helping to define the necessary quality criteria.” Prof. Hochreiter is regarded worldwide as a global pioneer of modern artificial intelligence and was instrumental in implementing Austria’s first degree program in Artificial Intelligence atJKU Linz two years ago. He is a board member of ELLIS, the network of excellence of the best European scientists in the field of machine learning an its applications. Hochreiter is also in charge of the LIT AI Lab and the ELLIS Unit Linz.
Putting AI through its paces
The first step along the way has now been reached: So-called supervised learning applications in low to medium risk areas are already being certified today. “We are already carrying out the first certification projects, whereby the applications are primarily to be found in an industrial environment, though also in the consumer sector. The next phases of the development cooperation will see an expansion of the current approaches, in order to be able to certify also safety-critical applications, based on a broader spectrum of machine learning methods”, says Dr. Stefan Haas, CEO of the TÜV AUSTRIA Group, and goes on to add: “We are particularly delighted to have the Machine Learning Institute of the JKU on our side as a highly competent and internationally recognized partner for this challenging venture.”
In the course of the certification, the machine learning models and their development process are investigated in detail on several levels. The partners not only investigate the actual function and reliability of the trained model, but also the safety of the software and whether this has been developed appropriately for the requirements of its field of use. And they check whether personal data is handled confidentially and whether possible ethical issues have been taken into due account.
Out now: White Paper (published by TÜV AUSTRIA and JKU): https://www.tuv.at/loesungen/digital-services/trusted-ai/
Institute of Machine Learning at the Johannes Kepler University Linz: https://www.jku.at/en/institute-for-machine-learning/
European Laboratory for Learning and Intelligent Systems: https://www.ellis.eu
TÜV AUSTRIA Group | Thomas Doms | TÜV AUSTRIA-Platz 1, 2345 Brunn am Gebirge
Email: digitalservices(at)tuv.at | T: +43 664 604546313
JKU, Institute for Machine Learning | Dr. Bernhard Nessler | E: Nessler(at)ml.jku.at | T: +43 664 3336009