Policy decoded: The Machinery Directive and AI

24 October 2019

In our new ‘Policy Decoded’ series, we break down a policy issue currently in the spotlight. We explain why it matters for the technology industries Orgalim represents. And we look at how we are working with our members across Europe and with EU policymakers to shape an enabling framework for the future.

In focus this time: how developments in artificial intelligence have put the Machinery Directive back on the policy radar - and what this means for the companies that rely on this important piece of legislation.

Scroll down to understand why the issue is in the spotlight, why it matters for the technology industries, what Orgalim is doing and how you can find more information on the topic.

What’s the issue?

The Machinery Directive is the core piece of EU legislation for the mechanical engineering industry: it promotes the free movement of machinery within the Internal Market, while setting out the ‘Essential Health and Safety Requirements’ to be observed when placing a machine on the market for the first time. The mechanical engineering companies Orgalim represents really value this Directive: in addition to ensuring a high level of safety for employees and end users, it has provided a stable legal environment for businesses since it came into force over a decade ago. Particularly for small and medium sized enterprises (SMEs), this stability is invaluable.

In the last year or two, however, the European Commission has been exploring the possibility of reviewing the Directive – prompted primarily by new developments in technologies like artificial intelligence (AI).

 

Why is it in the spotlight?

AI has climbed up the political agenda in recent years, reflecting public concerns that this technology is somehow different; that, in the most dramatic scenarios, systems deploying AI could start to act autonomously and provoke new, unmanageable dangers. The reality in industry is far from science fiction, however: AI has been deployed safely in manufacturing for many years. In fact, the European tech companies we represent have carved out a leading niche in the area of ‘embedded AI’ – AI applications integrated in consumer and industrial products.

Given the broader public debate around AI, however, the Commission has decided to take another look at the Machinery Directive to evaluate whether it is still fit for purpose when applied to machinery incorporating these new technologies. Currently, this evaluation is at the ‘impact assessment’ stage, with the final report expected in February 2020.

 

Why does it matter?

For the industries we represent, the Commission’s evaluation of the Machinery Directive is important for two reasons. First because the Directive itself is such a cornerstone of our companies’ legal environment, meaning a revision would cause much disruption and uncertainty. And second – zooming out – because a stable regulatory framework is essential for attracting investment precisely in new technologies like AI. Embedded industrial AI applications are among European industry’s world-leading strengths, and have the potential to boost our global competitiveness while enabling solutions to critical societal challenges. An enabling framework can help keep innovation and investment here in the EU in the face of fierce global competition.

 

What is Orgalim doing?

When it comes to the Machinery Directive, Orgalim has a primary role in engagement with the EU institutions as the main stakeholder representative for the industries working with the legislation. We act as an interface between our members from across Europe and policymakers in the Commission to ensure that the Directive continues to support the marketing of safe, innovation-friendly and market-relevant machines.

As the public and policy debates around AI have grown, Orgalim’s group of experts has been meeting on a regular basis to look in detail at how AI is used in applications where the Directive would apply. This has meant taking a deep dive into the fundamentals of AI: what exactly does it mean in an industrial context? What are the learning and decision-making techniques used in industrial AI algorithms? Is there really a danger of industrial AI beginning to ‘think for itself’ and take actions beyond its intended scope that could be harmful to humans?

The conclusions of these discussions were put forward in an Orgalim position paper in February 2019: ‘Orgalim comments on the upcoming Impact Assessment of the Machinery Directive’ (see below). Our industry experts were united in their view that industrial AI has been deployed effectively and safely in manufacturing for a number of years – and they see no reason or evidence backed up by the current state of technology development to suggest that this will change in the future. 

Equipped with this consensus and our experts’ insights, we have since put forward our views in meetings with Member States, participated in the Commission’s public consultation and met with the consultant tasked with the impact assessment. Now we are busy engaging with companies working with the legislation day to day to gather real-world examples of how they put the Directive’s requirements into practice for machines embedding new technologies like AI.

 

How can I find out more?

In addition to our views on the Machinery Directive revision, the Orgalim position paper takes a detailed look at what AI is and how it is used in our industries – covering questions like:

- How should we define AI based on how it is used in practice today?

- What are the boundaries of AI when used in industrial applications? What is ‘narrow AI’ and how does it compare to ‘general’ or ‘super AI’?

- What is machine learning and how does it relate to AI? What is the difference between ‘supervised’ and ‘unsupervised’ machine learning?

- How much control does the human operator have over machinery embedding AI functions?

Plus, our experts provide a sentence-by-sentence breakdown of the main articles of the Directive’s Annex 1 – the ‘Essential Health and Safety Requirements’ – to demonstrate how the existing text is equally applicable to new machinery incorporating AI elements.

For more information, contact Eleonora Piccinni, Manager of Technical Regulation, at eleonora.piccinni@orgalim.eu