KW 39: Zalando loses leading AI researcher, AI cannot be the inventor of a patent, appeals court rules, AI doctor: Neural networks learn from X-rays


Zalando loses leading AI researcher: Online retailer Zalando has confirmed that leading AI researcher Ralf Herbrich left the company at his own request at the end of August. The 46-year-old had been head of data analytics and machine learning at Zalando since the beginning of last year. With more than 50 patents of his own and numerous scientific publications, Herbrich is considered one of the leading minds for machine learning in Europe.

AI cannot be the inventor of a patent, appeals court rules: Artificial intelligence (AI) cannot be the inventor of new patents, the UK Court of Appeal has ruled. The appeal court ruled against Stephen Thaler, creator of a system called Dabus, who took a case against the UK’s Intellectual Property Office (IPO) which refused patents to his AI. It is the latest such judgement in a long-running battle to grant machines the status of inventor.

AI doctor: Neural networks learn from X-rays: In a collaborative effort, an international team of researchers has developed a deep-learning model that reads ethnic ancestry from medical X-rays with 95% accuracy. The researchers from the US, Canada, Australia and Taiwan found that the AI succeeded in this determination, using images of all anatomical regions of the human body. Already in the preface of their study, the researchers warn that this effect poses dangers for the automated evaluation of medical images and for AI-supported treatment suggestions.

– Advertisement –
Blockchain – Ticker – With our weekly newsletter we provide you with the most important developments in blockchain technology and scene.

OpenAI unveils model that can summarize books of any length: OpenAI has developed an AI model that can summarize books of arbitrary length. A fine-tuned version of the research lab’s GPT-3, the model works by first summarizing small sections of a book and then summarizing those summaries into higher-level summaries, following a paradigm OpenAI calls “recursive task decomposition.”

AI learning opportunities in medicine are expandable: A current study by the Institute for Medical Informatics at the Charité in Berlin illustrates the growing need for AI learning opportunities for physicians and hospitals. According to the study, AI skills are essential for medical practice, but are not being taught across the board in medical education, training and continuing education. According to the study, the majority of the medical training institutions investigated in Germany do offer students AI-related events, but mostly as elective courses or extracurricular activities.

This is how AI helps recruit the best talent: AI can help bring together promising job candidates and recruiting companies in a way that maximizes the benefits for both. The range of services can range from CV parsing and optimized job ads that target professionals in specific industries, to chatbots and video analysis of word choice, facial expressions and gestures.

Banks: How AI can support financial institutions in the compliance area
Security: How cybercriminals use AI for their purposes
EU AI regulation: Free ride for lobbyists
Hype Cycle for Artificial Intelligence, 2021: Four trends for the near future of AI
Knowledge platform: Artificial Intelligence simply explained

– Advertisement –
IoT – Ticker -The physical world meets the digital one. Internet of Things as an interface that revolutionizes both the industry and everyday life. Get a weekly update from the world of „Internet of Things“.


By 2026, a new computer center will be built in Leipzig for research into artificial intelligence and projects for its practical application. Germany’s federal and state governments want to provide almost 50 million euros.


Better than doctors? Machines are now supposed to help in the fight against breast cancer: The start-up „Hippo AI“ from Berlin wants to make artificial intelligence in medicine a common good. One major goal is to prevent large corporations like Google from taking over the market leadership in this area as well. Bart de Witte, head of Hippo AI, is planning one of the largest mammography databases. Hippo Foundation cooperates worldwide with clinics and research institutions that store the data, anonymize it and make it analyzable. These in turn have already offered the foundation several data sets, which will now all be combined into one package. Subsequently, the data is to be made available in processed form free of charge for research and development. Companies or researchers, in turn, commit to disclosing a resulting AI model and thus making it available to the general public. De Witte wants to finance his project exclusively with donations.

Amazon’s AI-controlled cameras will punish their drivers if they look in side view mirrors or are cut off by other cars, according to a report: Amazon’s AI-controlled cameras are designed to punish drivers if they make mistakes while driving, such as driving without a seat belt. At the same time, they will also be penalized if they look in their side view mirrors or are cut off by other cars. Penalties like these, according to the report, harmed the performance scores of employees as well as drivers. The declining performance scores reduced their chances of getting pay raises. Objections to false assessments by the AI are ignored. According to its own data, Amazon has seen a decrease in accidents since the e-commerce giant installed the cameras in more than half of its delivery vehicles.


AI recognizes destroyed houses after disasters: Researchers at Stanford University and California Polytechnic State University have developed the AI model „DamageMap“, an image evaluation system that, based on artificial intelligence and after extensive learning training, can use aerial photographs to make a damage assessment of buildings after natural disasters. Instead of looking for differences between before and after images, DamageMap uses photos of any kind taken before the fire to map the area and locate building sites. Then the program analyzes the images after a forest fire to identify damage through features such as blackened surfaces, destroyed roofs or the absence of structures. Like a human, the system can quickly identify whether a building is damaged or not without having to compare it to an image taken before the fire.


„Our program only needs about ten to 15 images to recognize deviating patterns. Usually, artificial intelligence neural networks have to be fed with thousands of data to achieve this effect.“
Denkweit CEO and co-founder Dominik Lausch, whose start-up checks e-car batteries with AI.


Using AI to fight legionella infections: Researchers of the company Legio Tools GmbH have developed an analysis system with artificial intelligence for drinking water monitoring. The system detects the bacterium Legionella, as well as bacteria and other contaminants in drinking water, with pattern recognition. The system can be installed directly in the house connection. Water enters the analysis system at regular intervals via appropriate regulating valves. The analysis system consists of a microscope with an image processor and other connected optical sensors. The sample is scanned, a neural network evaluates the scanned images and, thanks to corresponding training data, can precisely recognize whether it is Legionella, other bacteria, microplastics or sediment contamination.

Newsletter subscription

Subscribe to our free weekly newsletter for a compact overview of Artificial Intelligence topics:


More digital news briefings

Our political briefings