Automation, data, artificial intelligence… Industrial companies now have tools capable of guiding, or even directly producing, operational decisions. These systems are increasingly used to improve performance, secure processes, and facilitate trade-offs. The emergence of Industry 5.0 also raises questions about control over decisions and accountability.
Industry 5.0: what industrial companies have really changed in the way they manage performance
In 2025, the majority of industrial companies have integrated task automation, data, and artificial intelligence into their business processes. Only 6% say they are still at the very beginning of their digital transformation.
However, among those that have already launched digital initiatives, 41% consider the return on investment to be insufficient, revealing a growing tension between performance expectations and operational reality.
These findings come from the Industry 5.0 Barometer, published in October 2025 by the consulting firm Wavestone, based on a Franco-German panel of more than 600 industrial decision-makers.
The results reflect a shift in how industrial performance is managed. Beyond productivity and operational efficiency, digital projects are also assessed based on their ability to deliver decisions that are understandable and applicable in the field.
The issue is no longer solely about tools, but about the framework within which automated decisions are made.
Automated decisions: what choices are we actually talking about in industry?
In industry, automated decision-making applies to a range of situations in which digital systems generate, based on data, recommendations or trade-offs.
These decisions may involve production planning, task scheduling, equipment maintenance, inventory management, quality, or logistics.
In most cases, automation does not eliminate human intervention. Tools process large volumes of data, identify scenarios, or propose options, but the final decision is often validated by an operational manager. This assisted recommendation logic is now the most widespread, particularly when financial, human, or industrial stakes are high.
In other situations, it is possible to go further. Some systems automatically trigger actions, for example by adjusting production parameters, reorganizing flows, or launching maintenance operations. These decisions, made without the immediate intervention of an operator or manager, rely on models or algorithms defined upstream. It is these choices that are now raising questions for industrial companies—not because of their technical effectiveness, but because they shift responsibility for the decision.
When decisions slip out of our hands: the operational limits of automation
When decisions are produced by automated systems, the main challenge lies in how they are perceived and integrated within the organization.
When a scheduling adjustment, parameter change, or automatic trigger occurs without a clear explanation, the decision is executed but not fully understood. Operators and managers may then struggle to take responsibility for the consequences.
These situations raise the issue of accountability. In the event of an error, malfunction, or outcome that does not meet expectations, identifying the decision level becomes more complex. Responsibility is shared between system designers, business teams, hierarchical managers, and the tools themselves, without always being clearly defined.
Finally, automation can widen the gap between model logic and operational reality. Automated decisions do not always take into account unforeseen situations, local constraints, or weak signals known to field teams.
How can automated decisions be supervised in industry?
As automated decisions become more prevalent in industry, teams must understand the conditions under which a decision was generated, the parameters involved, and the possible margins for adjustment. Supervision thus becomes a role in its own right.
In practical terms, this supervision relies on the ability to identify situations in which the automated decision can be applied as is, and those that require human arbitration.
Teams must be able to interpret the recommendations produced by systems, taking into account production constraints, on-the-ground uncertainties, or current priorities.
This evolution changes managerial responsibilities. The decision-maker must assume final responsibility, even when decisions are based on automated tools. This requires a sufficient understanding of the mechanisms at work, as well as the ability to explain the decision to the teams concerned, to ensure acceptance and effective implementation.
Responsibility does not disappear with automation: it changes form and must be explicitly assumed. This is where discussions between industrial players take on their full meaning. Sharing feedback and comparing practices help identify effective decision-making frameworks and prevent automation from becoming a weak link within the organization.