As language assistants in the living room, in search engines on the Internet or as systems for autonomous driving in cars: Artificial Intelligence (AI) is already being used in many different ways and has long been part of our everyday lives. AI is no longer just fiction or the technology of tomorrow, but it has yet developed into one of the most important basic technologies today. We have already presented an overview of the origins and a definition of AI in one of our previous posts. In this article we show the universal applicability of AI by means of illustrative practical examples in different industries.
AI systems are used in more and more medical areas, for example in the evaluation of medical images. AI is used to identify and classify diseases and to make very precise diagnoses1,2. Another recent example is the development of a mobile application that uses the sound of coughs from the user to detect whether the coughing person has an (asymptomatic) case of Covid-193. With an accuracy of at least 98.5% in detecting infected persons, this method is very reliable. Another AI application can be found at the German Cancer Research Center and Heidelberg University Hospital. Here AI is used very successfully as a digital assistant to detect black skin cancer4.
In addition to these applications, AI is also used for administrative tasks in medical financial management and to improve the operational workflow in hospitals, for example in the organization of patient data2.
Law and Police
AI is also already being used successfully in the legal field. AI finds practical application in a software that searches databases with past jurisdictions and can then use them as the basis for current cases. Bryter develops a system where users can get answers from the computer to their specific questions, for example whether an item is tax-deductible5.
Police authorities are also increasingly relying on the use of AI. For example, AI software was used that independently searched and classified image material6.
Although the development of AI-based systems for autonomous driving is considered very demanding due to the high complexity and unpredictability of road traffic events, first applications are already in use7. The way to complete autonomous driving is divided into 5 stages, which differ in the degree of automation and thus also in their use of artificial intelligence. First applications such as automatic distance keeping and automatic lane assistants can be found on Level 1. On this level, AI is used in cooperation with sensors and cameras which enables shared control between the driver and the driver assistance systems (“hands on”). Level 2 is semi-automated driving. The vehicle can perform certain driving tasks such as lane keeping and distance keeping independently (“hands off”). For this purpose, the AI system requires significantly more computing power to perform all tasks simultaneously, which turns the vehicle into a rolling computer. Automated driving (Level 3) requires only limited driver intervention (“eyes off”), while fully automated driving (Level 4) and autonomous driving (Level 5) require no driver attention at all (“eyes off”, or “steering wheel optional”)8.
Image9: Test vehicle with sensor technology for autonomous driving.
AI is widely implemented in industrial production, especially in predictive maintenance, demand planning and product quality control10.
The automobile manufacturer General Motors (GM) has developed a system that can detect manufacturing errors during production with industrial robots even before the errors occur. Costs for unplanned production stops, which for GM sometimes amount up to 20,000 US dollars per minute of downtime, can be avoided10, 11. At the French food company Danone, AI is used in demand planning to anticipate demand fluctuations with the help of machine learning. The use of artificial intelligence reduces forecasting errors and prevents sales losses12.
Another application example of AI-based automated quality control can be found at the tire manufacturer Bridgestone, where product uniformity, i.e., the consistent quality of products, has been significantly improved thanks to an AI-based control system10, 13.
AI can support teachers by automatically grading students’ work and thus avoiding subjective assessments. The use of AI is also suitable for the targeted promotion of certain skills and the recognition of personal weaknesses of students. Some German schools use learning platforms for this purpose, which have a huge data pool of solved tasks and, based on the individual answers, make suggestions as to which tasks a student should continue with. In this way, knowledge gaps can be closed14.
Not only Internet giants such as Google, Amazon and Facebook exploit the user data and AI-based recommendation systems available to them to present individually tailored advertising15. AI-based systems also play an important role for other companies in targeting customers with personalized advertising. Companies hope to influence the purchasing behavior of customers and to design the purchasing experience according to their individual needs and wishes16.
With “Contextual Video Tagging”, for example, the German media house RTL uses an AI algorithm to link editorial content with matching commercials. In a cooking show, for example, when fresh ingredients are being used, the appropriate supermarket advertisement is shown17.
AI-based instruments detect and prevent fraud in the financial and insurance business. Machine learning models are also used to assess and process credit requests18. For example, Teradata has developed an AI-based platform that uses deep learning methods to detect fraud patterns in banking transactions, both in online banking and credit card transactions, as well as in branch and ATM transactions19, 20.
AI is also already being used in stock exchange trading. Investment banks, for example, use AI-supported trading bots that trade independently with financial products. This automated trading of financial products can save immense costs, since only faulty trades have to be checked by the employees and the complete personnel workload is no longer required for all successfully executed trades21.
1 Kobayashi et al. 2019: https://link.springer.com/article/10.1007/s11604-018-0793-5
2 Visvikis et al. 2019: https://link.springer.com/article/10.1007/s00259-019-04373-w
3 Massachusetts Institute of Technology 2020: https://news.mit.edu/2020/covid-19-cough-cellphone-detection-1029
4 Bundesministerium für Bildung und Forschung 2019: https://www.bmbf.de/de/was-ki-fuer-die-medizin-bedeutet-9177.html
5 Frankfurter Allgemeine Zeitung 2019: https://www.faz.net/aktuell/rhein-main/kuenstliche-intelligenz-fuer-juristische-problemen-16098581.html
6 Deutsche Welle 2020: https://www.dw.com/de/bergisch-gladbach-mit-ki-gegen-kindesmissbrauch/a-54636515
8 SAE International 2016: https://www.sae.org/standards/content/j3016_201609/
9 Shutterstock 2020: https://www.shutterstock.com/de/image-photo/fremont-ca-usa-feb-28-2020-1693673494
10 Capgemini 2019: https://www.capgemini.com/de-de/news/ki-in-der-industrie/
11 iFlexion 2020: https://www.iflexion.com/blog/machine-learning-image-classification
13 Harvard Business School 2018: https://digital.hbs.edu/platform-rctom/submission/bridgestone-production-system-innovation-through-machine-learning/
17 Handelsblatt 2020: https://www.handelsblatt.com/technik/digitale-revolution/digitale-revolution-werbung-im-perfekten-moment-wie-kuenstliche-intelligenz-eine-ganze-branche-veraendert/26308380.html?ticket=ST-4934412-9FcQtcQxEoFaUkhaSb4t-ap1
18 PWC 2020:https://www.pwc.de/de/finanzdienstleistungen/kuenstliche-intelligenz-im-finanzsektor.html
19 Teradata 2020: https://www.teradata.de/
20 Computerwoche 2018: https://www.computerwoche.de/a/ki-dienste-helfen-bei-der-betrugserkennung,3545722
21 Consorsbank 2018: https://wissen.consorsbank.de/t5/Blog/Wie-k%C3%BCnstliche-Intelligenz-den-B%C3%B6rsenhandel-ver%C3%A4ndern-k%C3%B6nnte/ba-p/75654
Author of this article is Robert Dehghan.