Letter IEDI n. 915–Revolution in process: OECD's overview of new technologies
In recent years, the development and combination of a set of technologies has increased the potential for profound transformations in production processes, management strategies and the characteristics of goods and services. The impacts will be transversal, reaching all economic activities, from agriculture to services to, obviously, the industry. Hence the frequency with which the term industry 4.0 or advanced manufacturing has been used.
The OECD's most recent Science, Technology and Innovation Outlook report gives an overview of the state of development, adoption and effects of these new technologies around the world. Today's Letter IEDI addresses this OECD assessment from two chapters of the report. Particular attention is given to Artificial Intelligence (AI), but other diffusing technologies —such as 3D printers, blockchain, nanotechnology and new materials— are also evaluated.
Under the title "Artificial Intelligence and the Technologies of the Next Production Revolution," Alister Nolan's study emphasizes the applications of these new technologies in industrial production and underscores the implications of specific public policies for the advancement and employment of each of them. Two aspects seem to be central to the design of national strategies for the future: how to increase access to high performance computing, whose demand will grow with the new industrial revolution, and how governments can support AI research and advanced computing.
According to the author, with the development of deep learning, AI can be applied to most industrial activities, from optimizing multi-machine systems to enhancing industrial research. In the latter, AI is already radically shortening the time required for the discovery of new industrial materials and the development of new products.
In addition to its direct uses in industrial production, the application of AI in logistics makes real-time management possible, significantly reducing costs. Also, this technology is being used to reduce energy consumption in data centers, assist in digital security, schedule meetings, manage expenses, and retrieve information and business data. Finally, AI is being combined with other technologies, such as virtual and augmented reality, to improve labor training and cognitive assistance.
Nonetheless, the development and diffusion of AI in the manufacturing industry is influenced by various types of policy and can therefore be facilitated or hindered. Policies that can have major impacts include: regulation of data privacy, which is crucial in the training of AI systems; rules of transparency and algorithmic accountability; research support measures; rules of intellectual property, among others.
According to Alister Nolan, governments, whose funding for basic research in universities and research centers in IA have been crucial, could also act favorably towards:
• Helping the development and sharing of quality data, acting as a catalyst and honest brokers
• Promoting open-data initiatives and ensuring that public data is disclosed in machine-readable formats for AI purposes;
• Supporting AI startups so that they have access to the computing power and technical expertise necessary for the commercial application of their ideas;
• Preventing the regulation of AI technology, whose impacts are still not well understood, to restrict innovations and their diffusion.
But if new technologies will revolutionize the industry, they have the same potential to revolutionize scientific knowledge itself. This is the topic of "Artificial intelligence and machine learning in science", by Ross King and Stephen Roberts, researchers of the University of Manchester and the Alan Turing Institute of the University of Oxford, respectively.
King and Roberts discuss the potential of AI technology and deep learning in broadening scientific knowledge and analyze current gaps in education and training that may compromise advances in the field.
In the authors' evaluation, the use of AI technology has the potential not only to increase productivity and reproducibility in science, but also to enable new discoveries that respond to the various global challenges of contemporary society, ranging from climate change to increasing antibiotic resistance.
Despite such advantages of complete "science automation," King and Roberts argue that it is the collaboration between human scientists and artificial intelligence systems that produces better science. While human scientists are still unparalleled in conditions that require flexibility and dealing with unexpected situations, algorithms process data on super-human scales.