Technology
Machine learning fits because compliance is full of uncertainty. There is no universal law of the case: the model is learned from that organisation’s data. It classifies, detects a pattern, points to a gap.
MIA
An intelligence that learns what we teach it.
MIA is a living intelligence, still at its dawn. It is born of ordering the norms, of years of practice in risk management, and of the wish to correct course from within while the boat keeps sailing.
An intelligence created from the ordered conceptualisation of the norms, from the knowledge that years of practice give, and from the wish to change course and correct errors, from within.
The journey has only just begun. The challenges ahead are immense beside what we have already learned, we and MIA herself. That is why we picture her young: dressed in green, with a gaze that is still learning and does not quite say whether it is human or synthetic.
She will learn what we teach her. We accompany her as she learns. She astonishes us, the way someone astonishes us whose talent has not finished growing. If she were a car, she would be a Formula One: the place where the latest advances are tested, always improving. MIA is to compliance what the leap to electric was to cars.
01
The scheme: from the norm to certainty
- 01NormThe obligations, ordered: norms, articles, impacts and measures.
- 02ModelThreat, vulnerability and impact, set into an equation.
- 03AlgorithmIt learns from each organisation’s data. It points to a pattern.
- 04JudgementA person, situated in a space and a time, validates and decides.
- 05CertaintyThe state in which legal risk is minimal.
02
The right use of artificial intelligence
Ethics
MIA learns what we teach her, and we accompany her. Two observers of the same event do not qualify it in the same way. The algorithm orders the file. The decision stays with whoever holds the judgement.
Norms
Everything starts from the norm and returns to it. Legal intelligence is collecting, evaluating, analysing and interpreting until data become useful knowledge. The destination is legal certainty.
03
The objectives, since 2015
In late 2015, with Universidad Complutense, the first team was formed to join mathematics and law. The question was specific: value a company’s non-compliance by its economic risk. The first frame was the Spanish Data Protection Agency, because it is cross-cutting and European.
Name the problem
Take compliance into a concrete business setting. Study the space between the company, the norm and predictive statistics.
Model
A probabilistic formulation. And another, when the data are not sharp, with fuzzy mathematics and aggregation operators.
Make it objective
Turn the subjectivity of “we comply” into a result that can be reviewed.
An intelligence that helps
Carry that procedure into software that works automatically, with the main purpose of helping organisations meet their duty to comply.
R(e) = Σ inc(n, e) · f(incum, pr)The risk of company e depends, for each norm n that applies to it, on the degree of non-compliance and on the probability of detection and sanction.
inc(n, e)It is 1 if the company must comply with that norm. 0 if the norm does not apply.
incum(n, e)Degree of non-compliance, on an ordered scale. It may be imprecise.
pr(n, e)The likelihood that the supervisory authority detects the breach and, where applicable, sanctions it.
04
The intelligence cycle
The roadmap set out in 2015 moved through several cycles. The last one is intelligence. That cycle is the starting moment of MIA. Along the way, each monograph consolidated a change in understanding.
- First team with the Complutense. Mathematics and law, on the same blackboard.
- Ecuaciones legales. How to predict future legal events.
- Compliance. A regulatory risk approach.
- The cyberlawyer. How to resolve the problems of the internet.
- MIA, still at her dawn. Permanent innovation as her reason for being.
The transformation was not only digital. It was organic, from the foundations to the understanding of compliance itself. The method of measurement is in Risk approach. The names and the calendar are in Origin.