Interest in understanding the influence of new technologies on the performance of our democracies continues to grow. It is good that the issue is attracting attention: it is of great importance to the evolution of digital democracy.<\/p>\n\n\n\n
For more than a decade, I have been drawing attention to this issue. In \u201cArgentina 4.0 The Citizens Revolution\u201d (Prometeo, 2013) I asked whether social media could help restore citizens\u2019 trust in politics. <\/p>\n\n\n\n
Considering recent publications, the question might now be reformulated as follows: can genuine digital democracy exist when our digital public sphere is mediated by mechanisms that we neither control nor fully understand?<\/p>\n\n\n\n
Conversation among citizens is increasingly ordered, selected, and partially rewritten by private systems over which we have limited control\u2014if any\u2014and whose operation we do not fully understand. Nor can their behavior always be explained or anticipated, even by those who design and govern them.<\/p>\n\n\n\n
In May 2025, I warned that artificial intelligence and social media could degrade democratic discourse. Since then, new research has begun to measure more precisely some of the mechanisms capable of producing that effect.<\/p>\n\n\n\n
The evidence does not show that artificial intelligence determines election outcomes, but it does show that algorithmic systems can structure political exposure asymmetrically, reward toxicity, modify certain attitudes, and even intervene imperceptibly in the way citizens express their opinions.<\/p>\n\n\n\n
Academic work published around May 2025 and, especially, over the following year allows us to draw several conclusions.<\/p>\n\n\n\n
The first is that there is still no solid evidence that artificial intelligence, on its own, has \u201cdestabilized\u201d democracies or produced massive and decisive shifts in citizens\u2019 voting behavior. Simon and Altay reviewed the available evidence on the influence of generative artificial intelligence in elections held during 2024 and concluded that its effects had been overstated.<\/p>\n\n\n\n
There was no \u201cmisinformation apocalypse\u201d of the kind that had been predicted. The authors emphasize the limitations of mass persuasion and political microtargeting, highlighting instead the influence of economic, cultural, social, and identity-related factors.<\/p>\n\n\n\n
The second conclusion is that platforms are not neutral channels. Algorithms organize visibility and can favor content, positions, accounts, or styles of communication. This is evident in studies of X\u2014formerly Twitter\u2014TikTok, chatbots, and AI-assisted writing tools. This does not necessarily imply an explicit ideological intention: it means that every recommendation architecture distributes attention, visibility, and opportunities for influence unequally.<\/p>\n\n\n\n
Third, algorithmic audits, observational studies, and field experiments allow us to conclude, with varying degrees of certainty, that engagement can degrade public debate. The evidence shows that partisan, hostile, or toxic content tends to attract more interaction, and that maximizing engagement does not necessarily satisfy users\u2019 considered preferences.<\/p>\n\n\n\n
A fourth conclusion is that digital literacy is necessary. Citizens need to understand that a conversational response, a writing suggestion, or a personalized feed is not a neutral representation of reality.<\/p>\n\n\n\n
Fifth, regulation and transparency are justified. The findings support the need for independent audits, researchers\u2019 access to data, explanations of recommendation criteria, and safeguards for tools that intervene in electoral processes.<\/p>\n\n\n\n
Sixth, taken together, the studies provide stronger grounds for arguing that the original promise of digital democracy is constrained. This is no longer merely a matter of opinion: the concern now rests on a more precise empirical basis. The capacity for interaction, decentralization, and access that inspired the original optimism now coexists with highly centralized private platforms that control the mechanisms of visibility.<\/p>\n\n\n\n
The results of these studies invite us to reject both panic and complacency. There is not enough evidence to claim that artificial intelligence and algorithms automatically control citizens or determine election outcomes on their own.<\/p>\n\n\n\n
But neither can they be regarded as politically neutral.<\/p>\n\n\n\n
The evidence shows that they can alter the information we receive, prioritize issues and positions, reward divisive content, and influence some political attitudes.<\/p>\n\n\n\n
The question, therefore, is no longer whether we should regulate, but how to do so without restricting freedom of expression or stifling innovation.<\/p>\n\n\n\n
If we want to prevent private, opaque, and automated decisions from ultimately governing public conversation, regulation of the algorithmic architecture of platforms is no longer optional.<\/p>\n\n\n\n
Regulation should focus on the processes, incentives, and architecture of these systems, rather than on policing content or opinions, and it should be proportionate to the risks posed by each platform or tool.<\/p>\n\n\n\n
At a minimum, it requires transparency, independent audits, researchers\u2019 access to data, traceability of recommendations, the option of chronological feeds, protection of personal data, and accountability for systematic bias.<\/p>\n\n\n\n
Regulation is not and must never become censorship: it should restore citizens\u2019 decision-making power and prevent digital democracy from quietly turning into a democracy governed by algorithms.<\/p>\n\n\n\n
<\/p>","protected":false},"excerpt":{"rendered":"
Interest in understanding the influence of new technologies on the performance of our democracies continues to grow. It is good that the issue is attracting attention: it is of great importance to the evolution of digital democracy. For more than a decade, I have been drawing attention to this issue. In \u201cArgentina 4.0 The Citizens Revolution\u201d […]<\/p>\n","protected":false},"author":1,"featured_media":2397,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"footnotes":""},"categories":[5],"tags":[],"class_list":["post-3056","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-digital-democracy-how-to-adapt"],"acf":[],"_links":{"self":[{"href":"https:\/\/carlosmagarinos.com\/en\/wp-json\/wp\/v2\/posts\/3056","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/carlosmagarinos.com\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/carlosmagarinos.com\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/carlosmagarinos.com\/en\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/carlosmagarinos.com\/en\/wp-json\/wp\/v2\/comments?post=3056"}],"version-history":[{"count":1,"href":"https:\/\/carlosmagarinos.com\/en\/wp-json\/wp\/v2\/posts\/3056\/revisions"}],"predecessor-version":[{"id":3057,"href":"https:\/\/carlosmagarinos.com\/en\/wp-json\/wp\/v2\/posts\/3056\/revisions\/3057"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/carlosmagarinos.com\/en\/wp-json\/wp\/v2\/media\/2397"}],"wp:attachment":[{"href":"https:\/\/carlosmagarinos.com\/en\/wp-json\/wp\/v2\/media?parent=3056"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/carlosmagarinos.com\/en\/wp-json\/wp\/v2\/categories?post=3056"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/carlosmagarinos.com\/en\/wp-json\/wp\/v2\/tags?post=3056"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}