Systems with an artificial brain, a robot as a work colleague, machines as gardeners: many possibilities of Artificial Intelligence are still a pipedream. But the technology is already changing our world. What Artificial Intelligence means and how far it has progressed today – a situation report.
According to a poll conducted by Bitkom in 2016, over 50% of farmers are already using digital technology in this country. Small wonder that sensors and robots in the barns and stables and GPS-controlled machinery, and drones on the soil and above it, open up a whole new world of possibilities, for a more efficient and resource-friendly, computerised agriculture. Four out of every ten farmers in Germany use digitally operated agricultural equipment. As far as robot technology in barns and stables is concerned, the share is about 40%. The farmers surveyed named savings in work and time, but also a better competitiveness, higher efficiency and less harm to the environment as main advantages.
percent of working Germans who would like an AI to support their superiors, according to a recent Bitkom survey.
trillion US dollars to be contributed by AI to the global economy by 2030 according to calculations. This would mean that the global gross domestic product (GDP) in 2030 would be 14 percent higher than without AI.
Basically, AI refers to the study of "intelligent" problem-solving behaviour and the creation of "intelligent" computer systems. These systems are considered intelligent when they solve tasks that humans can only solve with intelligent behaviour. Certain capabilities constitute the core of intelligent behaviour: perception, understanding, action and learning. Machines that only deliver results that have been programmed into them beforehand are not therefore considered intelligent. Rather, systems with AI must be able to put data into meaningful contexts themselves and learn from them. In order for machines to be able to do this, researchers virtually recreate synapses of the human brain, thereby creating artificial neural networks. Within these networks, for example, knight figures with AI move themselves in a computer game. Through their experiences during the course of the game, they learn on their own. With some practice, they can run to a fortress wall to defend it without human control or prior programming. Beyond cognitive performance, intelligence generally also refers to certain emotional and social abilities. However, only a few scientists are currently working on attempts to artificially imitate this kind of intelligence.
Man and machine have had a close relationship since the industrial revolution. With the introduction of Artificial Intelligence (AI), this relationship is taking on a whole new dimension. On the one hand, machines and technological applications simplify many everyday steps, but often they also appear opaque and threaten to replace people. According to a recent Bitkom survey, however, most Germans see more opportunities than dangers in Artificial Intelligence. For example, 62 percent of those surveyed stated that they did not feel threatened by the technology. Quite to the contrary, many hope that the AI will relieve their workload, particularly in the case of jobs such as time-consuming routine tasks. Another 44 percent of working Germans, for example, would like technology to support their superiors.
Unlike human intelligence, researchers basically divide Artificial Intelligence into two classes: strong and weak AIs. Strong AI describes systems that have a deeper understanding of problem solving and thus similar intellectual abilities to human beings, or that might even surpass them in the future. These include machines that think logically, learn independently and communicate in natural language – the prototype of a robot. In contrast, weak AIs generally refer to systems that focus on specific application problems and superficially emulate human intelligence. These include, voice assistants in smartphones, email spam filters and machine translation tools. The latter automatically compare successes and failures in translation and in doing so learn more and more sentence structures and linguistic contexts. So the more translations the software performs, the better it translates.
So far, only a few researchers have been occupied with strong AI. Most applications are based on weak AI and are used in a wide range of industries including industry, agriculture, office, medical, transportation and communications. The technology is still in its infancy – but according to calculations, Artificial Intelligence will contribute 15.7 trillion US dollars to the global economy by 2030. This would mean that the global gross domestic product (GDP) in 2030 would be 14 percent higher than without AI. In the European Union alone, the development of AI technology could increase GDP by around 2.7 trillion euros or 19 percentage points by 2030, according to a study by the McKinsey Global Institute.
One example where AI is already used extensively today is industry. More and more data from manufacturing and production processes is recorded in digital systems and displayed in real time. Intelligent machines can use this data, for example, to make predictions about the quality of the components they are currently producing. Cameras can take pictures of individual components and compare them with fault-free versions. The systems detect even the smallest deviations. If a component is damaged, a screw twisted or an enclosure scratched, intelligent machines detect the fault before humans can even suspect it. A message about the defective part is sent in real time, can improve product quality and save a lot of time and money, especially in mass production. Thanks to their Artificial Intelligence, the machines are also able to learn new things and detect errors ever more quickly.
Artificial Intelligence also helps to reduce machine outages in industry. Sensors, switches and intelligent tools capture, combine and analyse thousands of variables in production machines. Using this data, machine outages can be predicted before they occur, and repairs can be made proactively. Considerable progress: after all, it is usually unplanned outages that severely disrupt production schedules. This results in annual costs of around 50 billion US dollars in production. Machine maintenance using Artificial Intelligence can reduce downtime by 70 percent. Of course, this raises the question to what extent AI will replace people in the future. Researchers are also concerned with the socio-economic effects and consequences of the use of AI. The Organisation for Economic Cooperation and Development (OECD) has already convinced 42 countries to establish binding ethical rules on the development of learning robots, because intelligent machines should be developed on the basis of ethical principles for the benefit of all human beings.
AI is also used in agriculture, often in the form of machine learning. This refers to a subcategory of Artificial Intelligence in which systems can independently find solutions to problems on the basis of data and algorithms. An example: in indoor farming, sensors constantly measure all variables that are important for plant cultivation, including temperature, humidity, nutrient density and brightness. AI systems can use this data to determine the optimum conditions for plants. US researchers at the Massachusetts Institute of Technology (MIT) in Cambridge have, for example, planted basil in closed systems and fed all relevant plant data into an algorithm based on machine learning. From millions of different combinations, it found out the conditions under which basil develops its best taste. Including, among other things, uninterrupted irradiation with light.
So far, the use of AI for the cultivation of food has not been very widespread, but the first start-ups have launched AI-based systems on the market that make the cultivation of plants a fully automated process. The advantages: high-yield harvests, the resource-saving use of raw materials such as water and fertiliser and potentially also a more intensive taste experience. Experts believe that the use of AI could mean a quantum leap for global food security.
The examples from industry and agriculture show that Artificial Intelligence can accelerate and improve production processes. Despite its great potential, however, the areas of application of Artificial Intelligence are still limited today. An example: intelligent machines specialised in the detection of cancer cells are increasingly better at distinguishing cancer cells from healthy cells on the basis of a multitude of data using algorithms. However, these machines may not yet recognise the difference between cat and dog. Whether or when a more advanced Artificial Intelligence will be created that works on a variety of tasks and has the same social and emotional abilities as humans is still a hot topic of debate among AI experts.
