Ethical considerations of artificial intelligence
The revised ethics guidelines for “trustworthy” artificial intelligence were issued on 8 April 2019. Drawn up under the responsibility of a high-ranking group of experts closely associated with the artificial intelligence industry, it has been implemented by the European Commission. On the same day, the European Commission published its Communication COM(2019)168, in which it does not simply pass on the 38-page document drawn up by the experts to the European Parliament, the Council, the European Economic and Social Committee and the Committee of the Regions, but over eleven pages reiterates its own strategy, aims and plans. It only dedicates three pages to integration of the seven core requirements of the ethics guidelines, contextualising and amending them.
Tools in the service of people
The aim of the Commission is to “create trust in artificial intelligence that is aimed at people”. To this end it relies on ethical standards, controls and political decisions – and not on the marketing of artificial intelligence cloaked in trustworthiness to citizens who are overwhelmed and uninformed.
Bound by the values of the European Union, the Commission has taken charge of the publication and placed the emphasis on “ethical artificial intelligence”: “Artificial intelligence is not an end in itself, but an instrument that must serve people and should ultimately increase their welfare.” In contrast to this, the group of experts repeatedly and euphorically overestimates the new instrument, glossing over any possible risks. In addition, the group of experts stated in December 2018 that their ethics guidelines should not be used to stifle the development of artificial intelligence in Europe.
The officials of the European Commission amend the experts’ proposals subtly yet decisively, creating a transparent balance, whereby the potential of artificial intelligence is viewed critically and realistically as one possible way “to change our world for the better”. The associated challenges are taken just as seriously.
Amended perspectives of the ethics guidelines
The second major amendment by the officials relates to the perspectives of the ethics guidelines. While the 52-strong group of experts focuses mainly on economic development, international competition and above all, on artificial intelligence itself, the Commission concentrates its efforts on citizens and makes it clear that any trust placed in artificial intelligence has to be earned.
Accordingly, it reduces the annotations to the experts’ seven core requirements and reformulates them in line with its own perspective. It is worthwhile reading and analysing the differences in more detail, since the heads of the core requirements in both documents are the same: 1. Precedence of human activity and human supervision; 2. Technical robustness and security; 3. Private sphere and data quality management; 4. Transparency; 5. Diversity, non-discrimination and fairness; 6. Social and ecological well-being; 7. Accountability.
The difference between ethics and expertise
The fact that this loose fabric of open statements does not in itself give rise to ethical artificial intelligence is more than clear from the document by the group of experts. They were not able to agree on any red lines in the content that would make their ethical point of view clear: they were not able to achieve clarity in connection with citizen scoring, or clandestinely operating artificial intelligence systems! They were not even able to agree in their “e-guidelines” on their position regarding lethal autonomous weapons. In artificial intelligence, ethics and expertise are two different things.
For the time being, the work begun by the experts is now being moved on to the next stage. They are now being asked to submit their proposals for wide-ranging examination. Once these high-ranking experts have completed their task, by the beginning of 2020, “the Commission will evaluate the result (!) and propose the next steps” (cf. www.faz.net/aktuell/feuilleton/medien/kuenstliche-intelligenz-kommission-korrigiert-ethikleitlinien-16153875.html?premium)
It remains to be hoped that by then, politicians and the European Commission will replace the trust demanded by the economy and its experts in their approved artificial intelligence systems with citizens’ rights and ethical processes.