Traditionally, the EU goes on its summer recess until early September. One issue that has kept the representatives of the “Telecommunications Working Group” particularly busy—right up until the very last meeting—is the so-called AI Act ( Artificial Intelligence Act). The set of regulations, which is currently in the draft stage, is intended to establish a legal framework across the EU for all providers and professional users of artificial intelligence. This is reason enough for us, as developers and integrators of AI-driven robotics, to give this topic our full attention.
Is the Artificial Intelligence Act already “set in stone”?
In the spring of 2018, after three years of preparatory work, an initial proposal was drafted; however, it drew widespread criticism due to the very broad definition of AI. Since then, efforts have been made to arrive at a solution that—hopefully—takes all interests into account.
The latest proposals from Czech lawmakers seek to define AI more clearly and offer the following definition: “Autonomous systems that use algorithms and machine learning to achieve goals defined by humans.”—which closely resembles the way our robots approach tasks.
In our view, there is also significant room for improvement in the current risk-based approach (the higher the potential risk, the stricter the regulations), because rather than assuming a general risk, AI-based systems should only be considered a risk if they operate completely autonomously—that is, without human control, without human assessment of the situation, and thus without a final human decision.
Robotic AI + human intelligence = the perfect security partners
In virtually all safety concepts and operational scenarios for our robot portfolio, humans and machines are always used as parallel, complementary, and supportive components. Needless to say, robots such as SPOT, ARGUS or BEEHIVE Following your instructions, they patrol along predetermined routes, scan, check, and use their intelligence to, for example, autonomously avoid obstacles—and do so in increasingly efficient ways.
If incidents occur—such as the detection of an unauthorized person in the patrolled area—the robot immediately notifies the control center, where experienced human security personnel analyze and assess the situation and ultimately take appropriate action. This clearly illustrates the clear division of tasks as well as the limitations of AI. Or to put it another way:
“AI in robotics helps the system operate more efficiently, with optimized performance, and thus in a more time- and cost-effective manner, without rendering humans obsolete or relieving them of their responsibilities.”
AI-based systems therefore do not pose a potential threat to our industry, and this is precisely the message we need to convey. The further development of artificial intelligence must be given the space and guidelines it needs—it must not be constrained by a rigid bureaucratic framework.
Have a say, get involved, help make decisions
Innovative companies, developers, and AI experts should no longer view themselves merely as observers, but rather as stakeholders and experts alike, and should actively participate in shaping and implementing the Artificial Intelligence Act.
This gives us the opportunity not only to have a say in the international arena in the future, but also to make up ground and take on a leading role in the fields of robotics and AI. Let’s take advantage of them!