In the first half of 2024, Statistics Austria surveyed the use of information and communication technologies (ICT) in Austrian companies in line with the European data set. One in five companies already uses AI technologies, compared to one in ten in 2023. The strong growth is primarily attributable to the use of tools for text recognition and processing as well as speech generation.

The survey covered 6,600 Austrian companies across all sectors with ten or more employees, from construction and trade, manufacturing, energy and water supply, vehicle repair, transport, hotels and restaurants to ICT, real estate and services. “Proportionally, more large than small companies use AI technologies, and more companies from the service sector than from the manufacturing sector,” says Tobias Thomas, Director General of Statistics Austria. AI is most frequently used for text recognition and processing (65 percent of companies using AI), while speech generation (41 percent), speech recognition (29 percent), data analysis (around a third) and process automation (around a quarter) are also popular. AI technologies for image recognition and processing as well as autonomous driving machines or vehicles are less widely used.

And how are employees doing with it? Are they delighted or frightened when ChatGPT fluently writes down a text that could not be written so quickly with thought and the ten-finger system? And who checks afterwards what was generated? Is it factually correct or just good-sounding nonsense? Not to mention the fact that the miracle machine could have been trained in massive violation of copyrights. A former OpenAI developer told the New York Times in late October 2024 that the software company had used copyrighted content from the internet without permission. Since its release nearly two years ago, numerous artists, publishers and companies have sued OpenAI and other AI developers for alleged copyright infringement.

The free version of ChatGPT3 is the poster girl for AI use in everyday (working) life. The tool consistently does not use gender in German texts, which has to do with the fact that its training data is not free of discrimination and traditional role models. How good the output is depends on the version, how well you prompt – i.e. enter commands – and how much you understand the material yourself. Under these three conditions, it becomes clear how superficial the free result often is, how little differentiated or innovative the argumentation is, and how it also contains one or two hallucinations. That doesn’t matter to us when it comes to a birthday poem for grandpa. But not in other cases.

Studies on the world of work

Recent OECD surveys of employers and employees found that despite the trust that many workers have in their companies when it comes to introducing AI in the workplace, more could be done to boost confidence. Both sides are positive about the impact of AI on performance and working conditions. The concern about possible job losses remains. The surveys also show that professional development and more employee involvement would be desirable.

In a study conducted for the German Federal Ministry of Labor and Social Affairs in 2021/22, five scenarios for “Human-Technology Interaction 2030” were developed. The ambivalent scenario is considered to be the most likely. What is the ambivalence? Despite the growing performance and learning capacity of AI, the assumption for 2030 is that automation will remain trapped in the classic productivity paradigm. Real familiarity with artificial intelligence would be reserved for experts in innovation communities and open data usage would be restricted by limited access.

Automation solutions using AI could be used in industrial processes, human resources or controlling, for example. This would relieve employees of routine manual and cognitive tasks, which can be a sign of higher work quality. According to the study, the majority of employees in Germany would still not experience interaction with intelligent systems as an expansion of their scope for action and creativity in this scenario. Purely efficiency-oriented automation does not overcome departmental structures and continues to measure the potential of employees according to proven qualifications and approaches. Human-centered human-technology interaction could be designed differently.

© Siemens

Rebecca Johnson, a leading researcher at Siemens, is working on the further development of human-machine interaction. Her area of expertise: artificial empathy with the understanding that human-machine collaboration is not just a question of functionality. It is about concrete experiences in cooperation. How satisfied are users when interacting with digital companions? Can they communicate their intentions and can a bond develop? This bond is at the heart of Johnson and her team’s work. Together, they train avatars for digital interfaces and immersive environments, such as websites, digital displays or learning applications. As digital companions, they should be able to actively listen, understand and empathize: “We are well on the way to turning machines into our personal, digital and empathetic companions or assistants,” says Rebecca Johnson. Advances in infrastructure make this possible, says Johnson: “We have more powerful computers, higher bandwidths and better cameras to analyze human faces. We combine this with psychological insights and bring in specialists in psychology or negotiation to recognize what is important and evaluate it correctly.”

This brings us closer to the concrete wishes of Austrians for life in 2035. in 2024, the media agency dentsu surveyed 30,000 people in 27 European countries and 20 experts. 28 percent of respondents in Austria would like an AI clone of themselves to take care of everyday tasks, such as planning meetings or taking on purchasing, administrative and communication tasks. 60 percent want information to be seamlessly embedded in their own environment and surroundings through technology and to be accessible via devices. According to the dentsu survey, more than half of Austrians expect appliances and vehicles to reorder spare parts themselves and arrange service appointments independently.

© Getty Images

The OECD focused specifically on the financial sector and the manufacturing industry across all countries. Areas of application for AI tools included fraud detection, trading algorithms, chatbots in customer communication and tools for financial forecasting. In the manufacturing industry, AI is used for autonomous vehicles in warehouses, the visual inspection of objects or to detect potential inefficiencies. The influence of AI therefore affects a wide range of tasks and professions, but employment levels have remained stable. Overall, the use of AI in the world of work is more likely to lead to a reorganization of activities than to job losses. Automation reduces monotonous tasks. However, the introduction of AI technologies also requires a higher level of expertise and a broader range of skills.

Now is probably the right time to think again about which “sorcerer’s apprentices” are being launched and for what purpose. Artificial intelligence is a technology that can be used for a wide range of applications and tasks. What we can access for free via apps and browsers are mostly applications that five large technology companies have rolled out, developed and fed (Alphabet, Amazon, Apple, Meta and Microsoft). (European) in-house developments are sometimes better at building interfaces that actually keep people in the work process due to the stricter legal situation. More and more companies are adopting AI guidelines to regulate and signpost its use in order to create trust through transparency. Without expertise in the field of AI and the necessary digital literacy to use it sensibly and safely, it remains questionable how much the technology will help us. Because it also makes us vulnerable to manipulation or misinformation. And we wouldn’t be able to exploit their full potential because we might never develop the expertise we actually need.