Engineering Responsibility
What technical universities must now teach beyond code
The death of the neutral architect
Technology has long been taught as neutral. Engineers are trained to solve problems, optimise systems and build solutions that function reliably. The focus lies on precision, efficiency and performance. If something works, it is considered successful. For decades, that assumption held. But it no longer does.
Because the systems engineers build today do not remain within technical boundaries. They enter society. They shape behaviour, influence decisions and redistribute power. What once operated in the background now defines outcomes in the foreground.
The idea of the neutral architect is no longer sustainable. It is not evolving. It is ending.
From building systems to shaping reality
Artificial intelligence makes this shift explicit. What engineers design today does not simply execute instructions. It recommends, filters, prioritises and, increasingly, decides. Systems influence what information is seen, which options are presented and which paths are taken.
They do not just support decisions. They structure them.
The role of the engineer is therefore no longer limited to building functionality. It extends to shaping the conditions under which choices are made. And that changes the nature of the profession.
The question behind the system
This leads to a different set of questions. Not only: Can we build this? But: Should we? And more fundamentally: What kind of intelligence are we actually engineering?
Because every system reflects a series of decisions—about what is measured, what is optimised and what is ignored. These decisions are often embedded deep within technical processes, where they appear neutral.
They are not. They define how the system behaves.
The hidden layer of choices
Much of this remains invisible. Datasets are selected, often based on availability rather than representativeness. Models are trained on patterns that reflect past behaviour. Parameters are tuned to optimise specific outcomes.
Each step appears technical. But each step contains choices.
• Choosing a dataset is choosing a version of reality.
• Choosing a metric is choosing what matters.
• Choosing a threshold is choosing what counts as acceptable error.
Even the decision not to intervene—to accept default settings, to follow the path of least resistance—is itself a choice. And every choice has consequences.
Bias is not a bug
Bias is often framed as something to be removed. But bias is not an anomaly. It is inherent.
Every dataset carries historical patterns. Every model amplifies certain signals and suppresses others. The question is not whether systems are biased, but how that bias is understood, addressed and governed.
When bias is treated purely as a technical issue, its social implications disappear from view. And that is where the real risk begins.
From performance to impact
Engineering has traditionally been evaluated through performance.
Accuracy.
Speed.
Efficiency.
These remain important. But they are no longer sufficient.
Artificial intelligence operates in environments where impact matters more than performance. A system can be technically excellent and still produce outcomes that are undesirable, unfair or harmful.
It can score highly on accuracy, and still fail in practice. It can optimise processes, and still undermine trust.
The central question therefore shifts. Not: Does it work? But: What does it do?
The paradox of control
This introduces a deeper tension within the profession itself.
Engineers are trained to seek control—to understand systems, to predict behaviour and to reduce uncertainty. But many AI systems, particularly large-scale models, do not behave in fully predictable ways. They operate probabilistically, producing outputs that cannot always be traced back to a single, clear cause.
Responsibility, however, does not become probabilistic. It remains absolute.
This creates a paradox. Engineers are asked to take responsibility for systems they do not fully control. And that requires a different mindset.
Not less technical expertise. But a broader understanding of consequences.
The problem of accountability
As systems become more complex, responsibility becomes more diffuse.
Decisions are no longer the result of a single action, but of a chain of design choices, data inputs and organisational structures. When something goes wrong, it becomes difficult to identify where responsibility lies.
Yet the consequences remain concrete.
Someone is excluded.
Someone is misclassified.
Someone is affected in ways that are not immediately visible.
If responsibility cannot be traced, it risks disappearing into the system. And that is not a technical failure. It is an institutional one.
The role of the engineer
This changes what it means to be an engineer. No longer only a builder of systems. But a designer of consequences. A mediator between technical possibility and societal impact.
This does not reduce the importance of technical skill. It expands it.
Because understanding how a system works is no longer enough. It must be understood what that system does—within a broader social context.
Implications for technical education
Technical universities are at the centre of this shift.
They can no longer educate engineers as purely technical specialists, focused on optimisation and execution. The engineer of today operates in a landscape where technical decisions have social consequences from the moment they are deployed.
This requires a different educational model. Not one in which ethics is added at the end, but one in which responsibility is embedded from the beginning. Students must learn to think about systems not only in terms of functionality, but in terms of impact.
This means engaging with questions of bias, fairness and unintended consequences, not as abstract topics, but as integral parts of design.
Ethics is not a module
There is a tendency to treat responsibility as something that can be taught separately. A course in ethics. A lecture on societal impact.
But responsibility does not sit alongside engineering. It sits within it. It is present in every design decision, every dataset, every optimisation choice. To treat it as an add-on is to misunderstand its role.
Ethics is not what follows engineering. It is what shapes it from the start.
The future of the technical university
This also changes the nature of the technical university itself.
It is no longer sufficient to train specialists in code and computation. The technical university of the future becomes a hybrid institution, where engineering, social science and ethics intersect.
Not as separate disciplines. But within the same problem space. Where a line of code is understood not only as a technical instruction, but as a social intervention.
Europe’s position
In this transition, Europe occupies a distinctive position. It may not lead in scale or speed of technological development, but it has a strong tradition of regulation, public values and institutional accountability. This creates an opportunity.
Not to compete on capability alone. But to define the conditions under which capability is deployed.
Not everything that can be built should be built. And not everything that works should be used.
The deeper shift
The deeper shift is not technological. It is educational.
Engineering is no longer only about solving problems. It is about taking responsibility for the consequences of those solutions.
Students must learn not only how to build systems. But how to stand behind them.
What remains
So what remains of engineering? Not only code. But judgment.
The ability to anticipate impact, to recognise limits and to understand that technical decisions are also social decisions. Because systems do not simply function. They affect.
Final line
The question is no longer what we can engineer, but what we are willing to be responsible for.
This article is part of the series The University After AI, published in the Culture & Education section of Altair Media.
