Are AIs Essentially Collective Agents?
2023
This section explains the possible advantages and disadvantages of collective artificially intelligent agents.It introduces political issues, such as the importance of collective AI for democracy and the public sphere.This chapter expands on the conclusion of the previous one, namely, the claim that AIs may best be conceived of as collective epistemic agents.This possibility presents two immediate problems.First, in what sense are collective AIs autonomous epistemic agents?And second, how could we interact with these collective agents if they are so different in cognitive architecture and capacities?If it is already difficult for humans to interact with governments and corporations, why think that interacting with AI collectives would be less difficult, or a good idea in the first place?These questions only make sense if we assume that AIs are indeed collectives, somehow similar to governments and corporations in their complex data gathering, decision-making, and influence.If they are not, then we need to confront all the challenges about modeling AI in terms of individual human psychology, which were explored in detail in the previous chapters.It is because of the difficulties that individual human psychology presents for AI that conceiving them as collectives seems to be a good alternative, or at least one that deserves serious consideration.Thus, this chapter addresses these two key questions under the assumption that AIs are different not only because of their cognitive architecture, but also because of their collective nature.One development in AI regulation that already leans toward considering AI collectives is the legal protection of AI.Legal and political issues will be examined more carefully in the next chapter.The basic idea behind this proposal is that if AIs become important players in industry, science, and human affairs at large-as many expect-then protecting and regulating them under corporate 6 Are AIs Essentially Collective Agents?