While the farm machinery industry is busy trumpeting news of all the latest digital gadgetry it is bringing to market, the EU is equally busy sowing confusion over the legislation surrounding these developments.
In its latest position paper on the planned directives, CEMA – the association representing the European agricultural machinery industry – has pinpointed several problems with the proposals which it feels will be detrimental to the industry.
A smorgasbord of legislation
The first point raised is the absence of any strict definition of artificial intelligence (AI) and the term ‘safety function’, both of which are critical components of the directives.
There are two directives in the pipeline which cover the implementation of AI installed in farm machinery. The first is the general regulatory framework for AI, while the second is the machinery directive itself.
It should be noted that both cover all industries and not just agriculture, and that machinery is considered to be powered equipment fitted with a tool to carry out specific tasks.
This, as CEMA understands it, excludes tractors themselves which could well cause further confusion down the line as tractors and implements become more closely integrated, and that is before self propelled equipment is taken into account.
The question of AI and safety
CEMA notes that “the purpose of the future AI regulation is to protect individuals from the malfunctioning of artificial intelligence or its misuse for purposes that endanger safety or security”.
Thus the regulations are intended to cover all situations within society where AI is applied and will affect everything from toasters to autonomous vehicles.
However, there is proposed a secondary level of legislation which targets specific areas of activity, the Machinery Product Regulation being a case in point; it is the one which covers agricultural machinery.
The conflict arises through AI determining how the machine behaves and how that may change as the system gains experience, and critically, how it may affect safety functions.
This is seen as high risk AI by the umbrella legislation. Yet the Machinery Product Regulation accounts for this by requiring a full risk assessment before its implementation.
Lack of definition causing confusion
The situation is not helped because the EU fails to define exactly what it means by AI or safety function, yet liberally sprinkles the legislation with both terms.
If this were not enough, then further confusion is looming over proposed legislation to cover cybersecurity of radio equipment incorporated into machinery.
This aspect is already covered by the Machinery Product Regulation, so how will the two items of legislation come to accommodate each other? It is a question that has yet to be addressed.
Costing the digital revolution
On the surface, the advent of digitalisation promises many gains and advantages to agriculture. However, looking beneath the happy rhetoric of manufacturers, a viper’s nest of rules and regulations quickly becomes apparent.
As the technology becomes more complex, so must the rules governing it to protect loss or injury to the individual.
Conforming to the rules will not be cheap, and so we are faced with the spectre of any economic gains made through the deployment of AI being offset by the cost of its development and safe implementation.