By Kai Peters
Artificial intelligence ranks among the most controversial topics for lawmakers in Brussels. Some of them view AI as a major opportunity to help make the EU economy more competitive and improve the well-being of its citizens. Others regard the technology as a serious threat, one which puts millions of jobs at risk and endangers democracy, privacy or even the safety of Europeans.
There is however, an almost unanimous consensus that AI will have an impact on the lives and work of people in Europe and that, as a consequence, the EU must think about a political framework for its use. In April, the European Commission published a Communication on "Artificial Intelligence for Europe". As indicated by the title, the Commission is taking a rather optimistic view on AI. However, the text will also provoke lively discussions on the potential risks and the right ways to tackle them.
For the machinery industry, this debate is welcomed as it might help raise awareness, aid in understanding the implications of artificial intelligence (and Industrie 4.0 in general) better and support objective discussions. The reverse side to this is the latent risk that "Artificial Intelligence" is still misunderstood and that science-fiction movie scenarios will be conjured up, leading to fear-mongering, technophobia and backward-looking policy choices.
It is therefore of great importance to encourage a realistic debate in which the huge opportunities and challenges are not underestimated, nor are unrealistic expectations or unjustified fears created. As a starting point in this perspective, it is important not to confuse "Artificial Intelligence" and "Machine Learning" technologies. While "AI" as an overall concept conjures up the vague vision of a human-like, strong intelligence, the current reality is, and will remain for the foreseeable future, that the sub-area of "Machine-Learning" will be the driving force behind the technological change. However, despite the misleading title, the Communication’s approach poses the right key questions for a balanced debate.
Boost AI, boost the economy
The Communication addresses current challenges as well as potential problems that might occur. In detail, the Commission's proposal emphasizes three areas in which Europe would need to do more to promote or limit the development of AI: boosting uptake and investments, addressing socio-economic challenges and shaping the ethical and legal framework.
The most hands-on ideas certainly address the financial aspect of AI in Europe. New funds for research on AI have been proposed, as well as an investigation of how the European investment fund, the EFSI, could be used more effectively to finance AI-based innovations. It is very positive that the Communication explicitly highlights the huge potential of AI in industrial applications and aims to boost Europe’s industrial capacity and uptake across the economy, particular in light of the technological race against competitors in the US and China. The envisaged instruments - such as the AI-on-demand-platform, or a network of innovation hubs which is close-to-SMEs - are both promising and relevant.
Boosting the use of AI is a good idea, not just for some big software companies, but for the whole economy. Technologies such as pattern recognition, machine vision, data analytics and algorithms for predictive maintenance have already successfully been applied in industrial applications. It is expected that machine learning and the intelligent use of industrial data will contribute to further gains in competitiveness, for example by applying "human-like machine vision" for quality inspections, or optimizing the efficiency of complex production processes through adaptive, predictive control systems.
Use facts, see opportunities - not fears
In the same manner that technological changes have always had an impact on society, labor markets and job profiles, so too will AI, particularly as a part of digitization. However, before calling for drastic measures such as basic incomes and robot taxes, as some critics of AI do, it is important to be realistic about the short-term views and to be open to and prepared for future scenarios. VDMA has always argued against dystopian scenarios raising fears of unemployment and growing inequalities, stressing that countries with a high robot-density (such as Germany) produce more efficiently and can therefore manage to keep highly-paid jobs in the country. Some naysayers claim that "this time, it will be different".
It is very difficult to paint a precise picture of the future, but it is wrong to build future scenarios (or even policy choices) solely on a potential technical ability to copy human tasks. As McKinsey Global Institute points out in its report "Jobs lost, jobs gained: Workforce transition in a time of automation" (2017), other factors such as cost, acceptance and labor market conditions determine the use of automation. It is also wrong to look only at the jobs of today, as there is a good chance that new types of jobs and businesses will be created.
It is quite clear that job profiles will change, some of them dramatically, and that the workforce needs to be prepared to adapt to new requirements and even to change jobs. It is therefore essential to be ready: Employees by obtaining the right skills and education, companies by increasing their technological capabilities, and policymakers by updating educational systems and checking the labor market rules. It should also not be forgotten that increasing skills and availability of specialists might come too late to meet the growing demand for expertise in companies. It is therefore equally important to develop tools and instruments to lower barriers and to make the use of AI more efficient, for example, by designing low-barrier interfaces or using more "Self-Service" and "Guided Analytics" style approaches.
Automation and AI also bring important societal benefits to citizens and businesses. They can reduce strenuous and repetitive tasks and help increase capabilities, which in turn means there is more room for creativity and ingenuity. The systems can also help companies in the machinery sector lower their use of natural resources, stay competitive and continue employing people. Against this backdrop, the Commission has taken the right approach, namely one which balances a positive view of the technology with a rational analysis of the requirements of anticipating job transitions and economic patterns.
We will cross that bridge when we come to it - Solving future problems when they occur
It must be acknowledged that the Commission is on the right track in understanding that adopting AI technologies is not only about technology and financial aspects, but also comprises a legal framework. The AI package takes a cross-cutting approach and expands the view on topics such as data governance, liability and ethics. On these topics, the European Commission's Communication has initiated a debate on problems that will probably be relevant in the future. For example, it aims at drafting "AI ethical guidelines" and continuing the debate on implications of AI for liability regimes.
In VDMA's view, there is no doubt that the use of new technologies must always be in line with European values and fundamental rights. The basic principles for this, and for ensuring trust in technologies among citizens, are transparency and accountability. Particularly where far-reaching consequences for society, democracy and the rights of citizens are to be expected, simple, unexplained yes/no answers will not be sufficient to create trust in AI technologies. The implications of humans delegating decisions to machines in industrial settings must also be discussed.
However, ethical and legal constraints should not be determined too quickly and certainly not before differentiating the criticality of applications on one hand, and between short and long-term scenarios on the other. It should also be remembered that Europe has a functioning regulatory framework which addresses a wide range of questions raised by AI, whether it be safety (product liability, machine directive), transparency (GDPR) or security (the upcoming Cybersecurity Framework). Before rushing into new laws, careful assessment should be conducted to determine whether AI requires the wheel to be reinvented.
In current industrial applications, AI takes the form of machine learning to solve very specific problems which address precise technological or business questions where ethical consequences are still limited. Even in consumer applications, AI scenarios occur in quite distinct settings with limited degrees of autonomy. A smart kitchen machine, for example, might be able to learn how you like your meals best and take that into account when preparing the next dinner. However, AI will not make the kitchen machine develop a will of its own and start doing things that are completely unforeseeable. It is therefore important to ask how urgent ethical and legal questions really are in individual cases and to ensure that a premature discussion does not limit the scope for development of future technologies.
An intelligent start - for now
Fortunately, the Commission's proposal generally makes an intelligent start. "The EU can lead the way in developing and using AI for good and for all", the proposal emphasizes. It strikes a good balance by highlighting the benefits and by drafting a positive vision while providing a reasonable analysis of the risks and challenges.
There is, however, also a new role for the manufacturing industry. We must be aware that AI is not something that will only take place in factories and be discussed between engineers and business people. With the Communication, the EU has opened the debate on how to approach artificial intelligence from a EU political and societal perspective. It will be the task of the industry to convince people and politicians of the potential gains of these new technologies and thus eventually prove the pessimists wrong.