{"id":168,"date":"2021-09-01T20:10:37","date_gmt":"2021-09-01T20:10:37","guid":{"rendered":"https:\/\/lsi.asulaw.org\/softlaw\/?post_type=report&p=168"},"modified":"2021-10-18T16:32:16","modified_gmt":"2021-10-18T16:32:16","slug":"2-5-role-of-stakeholders","status":"publish","type":"report","link":"https:\/\/lsi.asulaw.org\/softlaw\/report\/2-results\/2-5-role-of-stakeholders\/","title":{"rendered":"2.5 Role of stakeholders"},"content":{"rendered":"\n
\n \n
\n \n\n
\n \n\n

<\/p>\n\n

2.5 Role of stakeholders<\/h1>\n <\/div>\n <\/div>\n\n \n<\/div>\n\n\n
\n
\n

Our analysis of stakeholders is based on the work of Abbott and Snidal, who created a heuristic device illustrating the distribution of stakeholders in a particular sector called the \u201cgovernance triangle\u201d [59], [60]. Through it, we observe three types of actors that led the creation or implementation of soft law programs (government, non-profits, and the private sector). The corners of the triangle contain unilateral relationships indicating that one organization leads a program. These are connected to bilateral relationships whose main attribute is the inclusion of two actors who join forces in a program\u2019s leadership. Lastly, the center of the triangle contains a category that combines all three actors.<\/p>\n\n\n\n

\"\"<\/figure><\/div>\n\n\n\n

Figure 3 is a representation of the governance triangle for programs in this database. What stands out is the preponderance of government as the generator of AI soft law with ~36% of all programs being principally led by a public authority. This is followed by multi-stakeholder initiatives (~21%) and non-profits with private sector alliances (~12%). One of the interesting dynamics in the database are the differences in the distribution of stakeholders in the top three positions in the country of origin category: the U.S., international, and Europe (see Table 8 below).<\/p>\n\n\n\n

\"\"<\/figure><\/div>\n\n\n\n

In the U.S., stakeholders enacting AI soft law are distributed among many groups, with the private sector and non-profits sharing their participation in the leadership with government institutions. A hypothesis to explain this phenomenon is that many of the leading firms in the sector come from the U.S. and, because of their global influence, they have taken an active role in the development of soft law. A different perspective emphasizes the reluctance of government entities in this country from participating in the management of AI, hence its parity with other sectors.<\/p>\n\n\n\n

Europe exhibits a picture unlike that of the U.S. Here, the government and its various alliances dominate the distribution of stakeholders with an 81% share, while the private sector by itself is marginally present with a 1.43% stake. Opposite to the U.S., European governments are known for adopting a proactive position in the governance of technologies, historically exhibiting a precautionary principle approach.<\/p>\n\n\n\n

Furthermore, it can be argued that a smaller proportion of leading AI firms are based in Europe, which may explain the lower levels of penetration by this sector.<\/p>\n\n\n\n

Finally, and far from surprising, the international front is led by multi-stakeholder efforts. Over half of programs are represented by all three sectors: government, the private sector and non-profits (e.g. standard-setting organizations, professional associations, among others). This is marginally followed by government, mainly in the form of multilateral initiatives.<\/p>\n\n\n\n

2.5.1 Government<\/h2>\n\n\n\n

An entity representing one of the three branches (judicial, legislative, or executive) of a public authority at any level (local, state, or national) is denominated as government in this database. Despite a definition of soft law that excludes the direct enforcement of government power, this type of organization unilaterally generated the highest proportion of programs in the database (~36%) through the enactment of non-binding instruments. If alliances with other types of organizations are counted, government is present in ~67% of our sample.<\/p>\n\n\n\n

To better understanding this dynamic, our team identified the level of government authority within all programs. As seen in Table 9, national and multilateral authorities are the most represented organizations in the leadership of soft law programs. In fact, over 78% of qualifying recommendations and strategies originate with government or one of its alliances, most of which were authored by national or multilateral authorities. This finding contradicts the popular narrative that soft law is simply industry self-regulation.<\/p>\n\n\n\n

2.5.2 Non-profit<\/h2>\n\n\n\n

Non-profits are organizations that do not distribute earnings amongst parties or officially represent a jurisdiction. Programs led solely by these entities correspond to ~9% of the database, a third originate from the U.S., and ~88% are created with the purpose of influencing external parties. Over two thirds of non-profit programs are either recommendations and strategies, which come in varieties such as institutional action plans [61], multi-stakeholder alliances [62], and principles [63], [64]. Interestingly, virtually all moratoriums or bans were promulgated by non-profits. Ten are related to autonomous weapon systems [54]\u2013[56] and one to AI-powered toys [57].<\/p>\n\n\n\n

2.5.3 Private sector<\/h2>\n\n\n\n

Approximately 11% of the database is composed of programs created solely by private sector organizations, half of which originate in the U.S. Firms had a strong preference for producing principles (~69%) and restricting their influence towards internal processes (63%) [65]\u2013[68]. Creating these programs provide organizations, especially large technology conglomerates with unique advantages. The creation of broad statements for the governance of AI is a relatively straightforward step that allows executive teams to broadcast their intentions with the technology to internal and external stakeholders. In addition, they can do so without necessarily having to develop and implement enforcement mechanisms that alter the organization\u2019s operation.<\/p>\n\n\n\n

Alternatively, private sector firms can choose to outsource the communication of their position with respect to AI by harnessing a special type of relationship classified in this research effort as alliances between non-profits and the private sector. Representing ~12% of the database, a large proportion of programs led by Np-Ps can be described as industry groups with significant ties to the private sector [69], [70] or professional associations that agglomerate and represent the views of industry professionals [71], [72].<\/p>\n\n\n\n

2.5.4 All sector alliance<\/h2>\n\n\n\n

Alliances between all sectors are the second most prevalent type of relationship, found in ~21% of the database. A large proportion (~45%) are standard setting organizations, who rely on representatives from all sectors of the economy to create technical specifications for a field. The second most popular alliance is spearheaded by governments who, in an effort to create inclusive AI strategies, invite non-profit and private sector representatives to co-author regional, national, and sectoral strategies [73]\u2013[75].<\/p>\n\n\n\n

2.5.5 Participants<\/h2>\n\n\n\n

Along with identifying the entities that represent a program\u2019s leadership, our research group distinguished organizations charged with a secondary role. Denominated as participants, they are characterized by their contributions in opining, discussing, or participating in a program\u2019s development. Figure 4 contains the distribution of stakeholders, which is limited to a quarter of the database (27%) since only that proportion of programs publicly indicated such roles. The largest group is represented by multi-stakeholder initiatives (~12%) such as government strategies [76].<\/p>\n\n\n\n

\"\"<\/figure><\/div>\n\n\n\n

<\/p>\n<\/div>\n\n\n\n

\n