The European Commission has recently proposed a new regulation that seeks to establish a legal framework for Artificial Intelligence (AI) in the European Union. And one of the most controversial provisions is the regulation of AI algorithms.
Algorithms are the essential part of AI systems, as they allow machines to learn from other data and perform tasks without human intervention. But on the other hand, they can generate serious problems due to their ability to reproduce and expand existing prejudices and inequalities in society.
That is why the European Commission has proposed a regulation that seeks to ensure that AI algorithms are developed and used in a responsible and ethical manner.
The proposal establishes that AI systems that present a high risk to people's health, safety or fundamental rights must undergo a prior evaluation and comply with a series of requirements, being transparent, explainable and auditable.
Transparency refers to the ability of developers and users of AI systems to understand how the algorithm works, and how a certain decision is reached. This involves providing information about the input data, the criteria used to make a decision, and the limitations of the algorithm.
Explainability refers to the ability of developers and users of AI systems to understand why a given decision was made, and how it was reached. And this means providing information about the intermediate steps that lead to a decision, and the factors that influence it.
Auditability refers to the ability of developers and users of AI systems to verify and validate the operation of the algorithm and its results. And this implies providing information on how the algorithm was evaluated.
But this is what the EU is trying to achieve, and what the companies that dominate AI technology are willing to concede is another thing, because I am convinced that they will resist tooth and nail to reveal their deepest secrets, and that limits are placed on their totally opaque big business today.
On the other hand, and as was to be expected, criticism has already appeared from companies that develop AI programs that do not wish to have any type of restriction in their work, and wish to keep their algorithms in the most absolute of secrets.
They always argue the same thing, that a regulation like this can limit the development and innovation of new AI systems. And what they do not want under any circumstances is that nobody controls what they do.
I believe that Europe has taken a very important step in the right direction, since regulation is necessary for a technology as important as AI to be used for the benefit of society in general and not only (as is the case now) for companies that master this technology.
We will have to wait for the results to find out if Europe has set any limits to the current lack of control of this new technology. And I hope it is so.
EN 
ES_ES 