The Human Factor Paradox
This thought started with something very small and entirely practical.
We are currently working on a complementary application for waste management, focused on collecting data in the field. In processes like these, one of the fundamental questions is always the same: how do you reduce human error as much as possible without removing the human being from the process altogether?
There is nothing particularly original about this as a technical principle. When information can be captured automatically, it is usually better not to ask someone to enter it manually. If the system can record the time, there is no reason for the user to declare it. If the location can be determined by the device, there is no need for someone to select it manually. If a photograph, an identifier, a scan, or another technical signal can confirm what happened, then these are safer than relying on the memory, judgment, or attention of a person who may be in a hurry, tired, distracted, or simply make a mistake.
The objective, therefore, is not to remove the human being. It is to reduce, as much as possible, the points at which the outcome depends entirely on them.
And somewhere along the way, I started thinking that this small technical problem might actually be a very good miniature of a much larger issue.
Because the human factor is almost always necessary. Human beings understand context, recognize exceptions, see what does not fit neatly into a rule, and can make a decision when data alone is not enough. At the same time, however, the human being is also the most unstable element in the equation.
A person can have all the evidence in front of them and still make the wrong decision. They can be influenced by emotion, pressure, personal interest, prejudice, a previous opinion in which they have already invested, a relationship, fear, or simply poor judgment. And often, the hardest part is not even making the wrong decision. It is accepting afterwards that it was wrong.
In everyday life, these things may be insignificant. A poor choice, a mistaken assessment, a decision we will not even remember the next day. Yet the exact same mechanism operates at a much larger scale. In business, management, politics, justice, investment, recruitment, and the evaluation of people and situations.
And at that level, the consequences are no longer small.
What is interesting is that as technology advances, more and more of what used to rely almost entirely on human judgment can now be supported by data. We can record, cross-check, compare, identify patterns, detect inconsistencies, and predict outcomes in ways that would have been impossible only a few years ago.
That is obviously positive. At least up to a point.
Because somewhere here a paradox begins to appear, and I think it will concern us far more in the years ahead.
We live in a world that wants, at least in its declarations, to become more humane. We constantly speak about justice, equality, meritocracy, transparency, reducing discrimination, and protecting the vulnerable. And rightly so.
But the more we demand that decisions be fully documented, consistent, and unaffected by subjective factors, the more we begin to turn against the very element that introduces that uncertainty.
And that element is the human being.
We do not even need extreme examples. We already see it around us. The more data we have, the harder it becomes to accept a decision based simply on “that was my judgment.” The more performance can be measured, the less room we leave for personal assessment. The more accurately we can predict an outcome, the more uncomfortable we become with a decision that deviates from it.
And this takes on a completely different weight when a human decision does not affect one person or one company, but millions of people. When it concerns public policy, legislation, geopolitical choices, economic models, wars, migration flows, public health, social welfare, or the way globalization itself is structured.
At that level, a mistaken assessment is not simply an error to be corrected in the next quarter. It can alter lives, create or deepen inequalities, move wealth from one part of society to another, and produce entire generations of people who feel that they are not participating equally in what we call progress.
And in a world where the sense of injustice is becoming increasingly intense, where inequalities in many cases grow rather than shrink, and where more and more decisions ultimately seem to be evaluated according to their economic return, our tolerance for human error becomes smaller.
From there, of course, the same mechanism moves downward into business and everyday life. A poor investment has a cost. A bad hire has a cost. A wrong risk assessment has a cost. A failure that could have been predicted is increasingly seen less as “human error” and more as poor management.
In such a world, the pressure for better decisions is entirely understandable.
The question is where that pressure may eventually lead us.
Because at some point, the question will almost inevitably be asked: if a system can evaluate more data, remain more consistent, never become tired, have no personal relationships, and remain unaffected by fear, anger, sympathy, or self-interest, why should the human being still have the final say?
And this, for me, is where the bigger picture begins.
I am concerned that, in our effort to create a fairer and more humane world, we may be building the conditions that lead us in exactly the opposite direction.
Not because some artificial intelligence system will decide that it no longer needs us. That is the easy science-fiction scenario.
The far more interesting, and perhaps more likely, scenario is that we ourselves will begin to consider human beings inadequate participants in important decisions.
And the argument will sound perfectly reasonable.
We will say that we want to reduce injustice. Minimize discrimination. Avoid mistakes. Protect the vulnerable. Make decisions based on data rather than personal judgment.
Yet if we follow that logic all the way to its conclusion, we may end up in a world where human judgment itself is treated as a problem that needs to be eliminated.
And that is where the real contradiction begins.
Because the human being is not only the source of error. The human being is also the source of empathy, forgiveness, second chances, and the ability to make an exception to the rule. A person can look at two cases that appear identical in the data and understand that, in reality, they are not.
Technology can tell us that someone failed.
A human being can understand why.
Technology can tell us which decision is the most efficient.
A human being can decide that it is not necessarily the right one.
Technology can calculate the cost of a choice.
A human being can understand that not everything of value has a price.
This is why I believe the right direction is neither to trust humans blindly nor to transfer responsibility to machines. It is to use technology to make human decisions better.
We need to design systems in which the machine makes it harder for the human being to make a mistake, without taking away their right to be human.
Systems that provide more evidence, reveal contradictions, point out when a decision deviates from what the data suggests, and, when necessary, require an explanation for choosing a different path.
But the final responsibility should remain human.
And perhaps this is far more important than it appears today.
Because the greatest challenge of the age ahead may not be how to make machines more human.
It may be how to avoid creating, in the name of a more humane world, a world in which the human being is ultimately seen as the weak link that needs to be removed.