Why does business need calculation

How artificial intelligence is revolutionizing the order of the economy

It has always been a dream of mankind to overcome the natural limits of man, to create a Deus ex Machina. The fantasies of artificial intelligence (AI) find very different forms and representations in cultural history. They range from utopias of material wealth and eternal life to dystopias of the subjugation of humans by supernatural and, in this sense, artificial intelligence.

Science has been looking at the subject more seriously since the 1960s. The initial euphoria gave way to disillusionment. Computing and storage capacities were too low to use artificial intelligence in practice. This has since changed. Thanks to a technological leap, we are now able to collect massive amounts of data (big data), store it and intelligently link it with one another using algorithms. So it is no coincidence that the topic of artificial intelligence is back at the top of the scientific agenda, but that business is also interested in practical applications. Beyond the diverse use cases, however, artificial intelligence will revolutionize the order of the economy and raise serious ethical, regulatory and social questions that we do not yet fully understand.

Artificial intelligence is a blind super brain

When we speak of artificial intelligence today, we mean very different forms, and occasionally cognitive systems are spoken of in more general terms. Artificial intelligence means that computers and machines develop cognitive skills such as “learning” and “problem solving” and, in this sense, are able to act autonomously. Narrow (weak) artificial intelligence is the term used to describe the ability to control defined and structured processes using artificial intelligence. The advantage is that these processes are carried out faster, more precisely and less prone to errors. Broad (Strong) Artificial Intelligence is understood to mean that artificial intelligence expresses an autonomous behavior that is also able to learn itself and thus develop independently. Machine learning and deep learning are subforms. Unsupervised learning, which independently recognizes patterns from unstructured data, enables a certain decoupling from human intelligence or a kind of co-evolution. The replication of neural structures up to the development of an artificial consciousness in the sense of self-reflection and idea-guided creativity are still relatively far away even today. In the long term, creative and context-sensitive abilities remain human talents. The development of artificial intelligence is not based solely on mathematical logic and information technology. Neuroscience, psychology or linguistics are also involved.

Data is of central importance in the application and development of AI. Without big data, there would be hardly any applications of AI. Conversely, big data is useless without AI, because it leaves large amounts of mere data from which no patterns can be recognized and decisions cannot be derived. Data have the economic peculiarity that the value of a single data set is practically zero and only the combination of as much data as possible on network effects generates the economic value. Only AI turns “big data” into “smart data”. At the same time, the danger of a confirmation bias must also be pointed out in this context: connections that have been identified - even those wrongly assumed - are replicated and confirm themselves. Data science is of crucial importance in this context. It extracts knowledge from data that is used to develop algorithms, which in turn contain a logical evaluation rule and decision instructions that autonomously control the handling of data. In turn, data is also the basis for permanent Bayesian updating of the algorithms through reinforced learning.

AI is the key to creating value from data

Digitization can be described as the technological possibility to use the added value in data economically. Using artificial intelligence to recognize patterns in data, control processes, make decisions and learn from new data, is the key to digitization alongside the data itself. The decomposition of “knowledge” shows how data can be used to create economic value and how AI can leverage this potential for value creation. The core of the economization of AI is that it can deal with large, even unstructured amounts of data faster and more systematically, i.e. more efficiently than humans. Technologically, the added value of the data is only accessible through AI and the realization of the added value becomes economically profitable. The use of AI can reduce production costs because processes can be controlled faster, more precisely and more reliably. The use of AI creates new products, especially in the area of ​​predictive and contextualized services. Ultimately, new knowledge and innovations can be generated systematically. AI thus represents a possibility to endogenize the growth and innovation process, which is sometimes described with the term singularity. In economic terms, AI is therefore not a continuation of industrial automation by other means, but a qualitative leap into the autonomization of processes and decisions.

AI makes data the third production factor

The simplest conceptual approach to the representation of the economic use of technologies is the "production function", which contains a rule for the technologically possible and economically efficient combination of resources. Macroeconomically, the development of the economy can be represented by the evolution of the production function. It ranges from the agricultural economy to the industrial economy to the future data economy. In this sense, AI can be understood as an extension of the technological possibilities to combine scarce resources with one another in order to produce more and also new output. As a result, data appear as an independent production factor in the production function alongside capital and labor. At the same time, this changes the substitution relationship between the production factors and thus the functional distribution of income between data, capital and labor (see Figure 1). Just as the economic rent accrued to the factor soil in agriculture, in industrial capitalism the factor capital gained market power over the factor labor. AI will now completely redefine the relationship between data, capital and labor.

illustration 1
Four interdependencies from adding data to the production function

Source: J. Schneider, M. Weis, H. Vöpel: Artificial Intelligence and the Reorganization of the Economy, 2018.

AI replaces specialized work with data-based algorithms

AI increases the substitutability of highly skilled work and specialized human capital. Knowledge and experience are no longer tied to people and no longer require any prior training. AI thus becomes unbound knowledge, which in the form of learning algorithms maps all competence and experience and makes it transferable. Tied up human capital, on the other hand, always requires training and an individual learning process. Both cause costs, which should be refinanced through the specialization advantages realized later. Specialization is not a natural human wish, but an economic (investment) calculation. AI can now raise and merge all knowledge and experience without significant specialization costs. The qualifying wage bonus for specialized work is decreasing as AI will increasingly also perform cognitive and non-routine activities. In the long term, there is no threat of mass unemployment through the use of AI, but the calculation of training and job profiles of occupational profiles will change massively.

AI combines capital with data

In addition to tangible assets such as machines and factories, the data is an intangible and replicable asset. The use of the data shows non-rivalry and enables cross-functional value creation. The logic of industrial production is primarily based on the realization of specialization advantages and economies of scale and results in standardized value chains based on the division of labor. This form of production has made it possible to continuously translate technical progress into a higher degree of automation. The result was economies of scale, because the fixed costs and thus the capital costs of a company became ever greater with specialization, the variable costs of mass production ever lower. In the future, however, the economic rent will fall more to the data than to the capital, which as a production factor will be less specific. This reduces the cost of capital. In the medium term, AI could thus reduce the interest rate and thus the capital income as a less intuitive effect in the sense of the capital and production reversal theory of Böhm-Bawerk. This transition means nothing less than the replacement of industrial capitalism by data capitalism. Just as capital occupied labor in industrial capitalism, so in data capitalism data will dominate capital because they siphon off economic rents like the land factor in agriculture and capital in industrialization

AI is redefining the relationship between capital and labor

The addition of data as a production factor through AI also changes the relationship between capital and labor. For certain, predominantly less qualified jobs, the ratio to capital is weakened, i. H. the work intensity decreases, e.g. B. in the transport sector, where capital is still needed, but hardly any labor. Conversely, it is much easier for new competitors to enter the market without capital, as the use of the data does not depend on capital. Examples are hotel or taxi brokerage platforms. The know-how of capital becomes less important compared to the know-what and the know-when of the data. The rapid and largely capitalless market entry lowers barriers to entry and exit, which increases the contestability of markets compared to the scalable network effects of platforms. Overall, the increasing dependency of labor on capital, which is increasing in some areas and decreasing in other areas, leads to a more uneven distribution of income for the labor factor.

AI is replacing the industrial with the digital order

The technological changes in the production function entail fundamental changes in the industrial organization of sectors and companies. The vertical industry structure will give way to a diagonal platform structure and data architecture. In the course of this, companies will transform themselves from specialized production units into hybrid and open collaborations. The value creation potential between companies and industries will be greater than that within industries and companies. Value chains are shaped by interfaces. They mark the optimal balance between specialization based on the division of labor and organizational synergies within companies. AI will enable vertical integration especially at the interfaces, i.e. the transitions between specialized competencies (see Figure 2).

Figure 2
AI order: from vertical value chains to diagonal data architectures

Source: J. Schneider, M. Weis, H. Vöpel: Artificial Intelligence and the Reorganization of the Economy, 2018.

AI is becoming a key ethical, regulatory and geopolitical issue

The revolution of the economic order in the data economy through AI requires a restructuring of society as a whole. This applies to digital ethics and competition law as well as education, the labor market and the welfare state. In addition, the development, control and application of artificial intelligence have long since become a geopolitical question of power, a war on big tech, as The Economist recently put it. Even the question of the interdependence of the order of the market economy and democracy is raised in connection with AI. Against this background, the development of a holistic AI strategy that includes questions of data sovereignty, ethics and regulation is necessary. For the German economy, which, like no other, is broken down into defined and delimited processes and is therefore predestined for the introduction of AI, AI is becoming the perhaps decisive question of prosperity and competitiveness.

  • 1 Cf. G. Dow: Why Capital Hires Labor: A Bargaining Perspective, in: American Economic Review, 83rd Jg. (1993), No. 1, pp. 118-134; and J. Haskel, S. Westlake: Capitalism without Capital, Princeton 2017.

* This article is based on the study by HWWI and Ernst & Young, see J. Schneider, M. Weis, H. Vöpel: Artificial Intelligence and the Reorganization of the Economy, 2018.