|
| An artificial intelligence robot statue as a judge. |
In 2024 tweet, Elon Musk stated that artificial intelligence (AI) "will soon beat doctors and lawyers." While his assertion may seem hasty and extreme, it is undeniable that AI is transforming multiple disciplines and is becoming an indispensable tool.
In this article, we will attempt to relate AI to one of the most complex and debated issues in criminal law: the distinction between dolus (intent) and culpa (negligence), which, in the words of the German legal scholar Hans Welzel, represents "one of the most difficult and controversial questions in criminal law."
This blurry boundary between dolus and culpa generates significant inequalities and arbitrariness, which often impact the proportionality of sentences and undermine the value of justice, the ultimate goal of criminal law.
The Difficulty in Determining Eventual Intent
Our criminal code lacks a definition of dolus, a fact that provides some degree of freedom to judges in choosing one of the definitions offered by legal doctrine. Such discretion conflicts with equality due to the sheer number and variety of theories, each with its own disparate consequences. The other issue is the procedural difficulty judges face in accessing the mind of the perpetrator, as it is within the mind that the existence or absence of dolus is determined.
Indeed, dolus is intention; more precisely, it is knowledge and willingness to commit a criminal act. However, in dolus eventualis (eventual intent), that knowledge and willingness are weaker; it is enough to show that the subject foresaw the result (knowledge) and either accepted it or was indifferent to its occurrence (will).
This indifference to the result, in terms of the subject's internal or emotional attitude, is very difficult to prove within a criminal trial, which leads to intuitive, discretionary, and sometimes unjust rulings, as they rely on what an average person should have known at the time of the crime, according to the rules of sound criticism, experience, and logic. In other words, dolus is generalized.
For this reason, some authors have proposed this generalization in order to achieve greater uniformity in its application. They argue that, rather than probing the internal attitude of a subject, we should focus directly on the behaviors or dangers that the legal system requires to be considered under threat of punishment for a criminal offense. That is, instead of examining the mind of the perpetrator, one should separate the dangers that warrant a finding of dolus from those that are classified as culpa.
This position, not so far removed from judicial reality—where dolus is often assigned based on what the subject "should have known or understood" rather than what they actually knew—will contribute to legal certainty, as it will allow us to know in advance which risks fit within dolus and which fall under criminal negligence.
The Role of AI in Determining Dolus
Sufficient danger and the probability of producing the result are elastic situations, and therefore manipulable by human beings. Providing them with content could be a simpler and more precise task for AI through pattern recognition, using various indicators. To give just one example:
1. Predictive Behavioral Analysis
Using machine learning models, one could evaluate countless prior cases and, after feeding it the data of the current case, determine and measure the concrete level of danger associated with the behavior.
2. Evaluation of Context and Background
AI could analyze previous decisions made by the subject, identifying whether they have exhibited a systematic attitude of indifference to similar risks.
3. Simulation Models
Through simulations, one could evaluate the level of risk the subject perceived and their attitude toward it.
In this vein, AI could establish more precise normative parameters, allowing for a more accurate evaluation of factors such as:
- The probability that the harmful result will occur.
- The subject's ability to foresee and avoid the risk.
- The severity and immediacy of the danger posed.
- The importance of the legal interest at risk.
- The controllability of the causal chain, etc.
Conclusion
Without a doubt, the application of AI in criminal law presents ethical and technical challenges, but it also offers an opportunity to reduce judicial discretion and contribute to more objective, reliable, and, above all, predictable and equitable justice.
While it is unlikely that AI will ever replace the judgment and reasoning of humans, which must underpin every criminal decision, its capacity to analyze vast volumes of data and detect patterns could become a key tool for judges and contribute to more predictable and uniform justice.
