Given the character of the European Ethics Guidelines for Trustworthy Al as a ‘living document’ and on the occasion of the First European AI Alliance Assembly organised by the European Commission (June 26, 2019), I would like to share with all of you the following:
The trustworthiness of Artificial Intelligence, which is vital in overcoming the resistance of local societies and ensuring the widespread acceptance of technological developments, is in my humble opinion dependent also on the following conditions:
- The strengthening of democratic processes and, in particular, the establishment of societal controls to ensure that these applications do not come about at the expense of the many by increasing the power of the few.
- The applications of Artificial Intelligence, instead of aspiring to uniformity at the European (and global) level, should be guided by local social customs and local moral attitudes so that individual societies do not reject them as alien and irrelevant. It is certainly a project that demands the European principle of ‘unity in diversity’.
- The average European citizen cannot enthusiastically welcome these applications as long as there is no strict definition of the degree of autonomy we are willing to accept for this type of systems. If the trustworthiness of these systems is to be guaranteed, there is urgent need for a European Commission initiative to create a legally binding framework, which would prevent the creation of sophisticated man-machines that would be flagrant violations of human dignity. Ultimately, the widespread acceptance of Artificial Intelligence systems depends on ensuring that the dignity of humanity cannot be placed at the greatest risk through its instrumentalisation and objectivisation.
- It is exactly this question that confirms the need for a European defence policy founded on our own cultural and moral system. If the United States of America and China proceed with the construction of man-machines for defensive purposes, why should we follow them by sacrificing the values of European civilization, rather than developing our own European defence system, not with man-machines but against man-machines?
- It is therefore judicious to exclude machines from the protection of individual rights, which are legally prescribed solely for the protection of human beings. The intention in certain circles to extend the protection of individual rights to include machines creates confusion in society regarding the distinction between man and machine, which in turn threatens a mass rejection of machines by society.
- In conclusion, we believe that the degree of social acceptance of Artificial Intelligence systems depends on the degree of self-restraint of all those involved in the design and commissioning of these systems. If the 19th century was the century of progress and the 20th century the century of autonomy, the 21st century must be the century of self-restraint. Modern man seems to find himself once again before the Tree of Knowledge. Can we be wise and cautious enough this time not to cut down its last fruit prematurely? If we do not restrain ourselves morally and legally today, those of our grandchildren who are not mechanically and biologically "modified" tomorrow will be second-class citizens in relation to sophisticated man-machines, with a mathematical certainty that they will share the fate of the native peoples of Australia and America. Do we have the right to permit such a possibility for our grandchildren and their children? And as long as we cannot exclude such a possibility, how can we expect intelligent human societies to embrace the new AI applications unreservedly? Since not only our progress and autonomy, but also our social coexistence and the security of our planet will ultimately depend on our respect for the principle of self-restraint, we propose the addition and indeed the prioritisation of the principle of self-restraint in the four basic principles of the European Ethics Guidelines for a Trustworthy Artificial Intelligence in its first upcoming revision.