In early June 2024, "Situational Awareness" by Leopold Aschenbrenner (Aschenbrenner 2024) was published, a short work that aims to show a future perspective in the short, medium, and long term on the evolution of artificial intelligences. The work, whose creator is one of the best-informed people in that industry, having worked in executive positions at OpenAI, is essential reading for understanding the current times and those to come, perhaps in a short space of time. Predicting the future is always risky, but in this case, Aschenbrenner starts from first-hand knowledge, and rarely does a person linked to the development of AI venture into the arena, issuing well-founded opinions and not a few warnings. However, the book, which covers many technical aspects of great interest, at times, especially in the final stretch, takes directions that can be considered worrying.

As an European with a markedly antimilitarist ideology, I offer a modestly critical and reflective vision of the clearly militaristic approach present in certain chapters of Aschenbrenner's book, above all, as previously mentioned, a reflection on the future and consequences of the expected evolution of all types of AI in the coming years, a work of great validity that raises scenarios that are both exciting and worrying. This perspective highlights a certain paramilitary and inevitably imperialist tendency on the part of Aschenbrenner, as well as a perception that we could describe as schizophrenic about the security and pre-war danger of the current geopolitical situation, which could worsen in the future with the advent of Artificial General Intelligence (AGI). This approach will be contrasted with the anti-colonial and post-colonial ideas of Boaventura de Sousa Santos in his work "Decolonizing Knowledge, Reinventing Power" (de Souza Santos 2010), and the worrying implications of a US-centered AI for Europe and other countries in the American continent will be analyzed.

Aschenbrenner's insistence on preparing (his society and nation) for a potentially conflictive future after the eventual emergence of AGI reveals a stance that not only prioritizes national security in a way that seems excessive but also appears to justify the militarization of research and development of AGI itself. This attitude not only perpetuates a vision of the world dominated by competition and confrontation, but also reflects a post-democratic mentality, of walls and iron curtains, where decision-making shifts from democratic processes to militarized and technocratic institutions.

Boaventura de Sousa Santos, in "Decolonizing Knowledge, Reinventing Power" (de Souza Santos 2010), offers a perspective that markedly contrasts with that of someone like Aschenbrenner, who not only belongs to a later generation but inhabits a world perhaps different from that of 14 years ago, and in which technology and ideology can sometimes merge into a certain obsession with security. Santos advocates for a decolonization of knowledge and a reinvention of power that is more inclusive and democratic. His approach rejects the hegemonic and colonialist power structures that have historically dominated knowledge and decision-making. Applying this perspective to the analysis of Aschenbrenner's book, it can be argued that his militaristic and seemingly Pre-war approach appears to perpetuate (undoubtedly without full awareness of it, each person is formed in their zeitgeist) a certain form of what we could christen as technologized epistemological colonialism. By focusing security and defense on militarization and control of information (this second aspect, common in the business world, where secrecy and patents dominate the possibility of obtaining new markets), Aschenbrenner ignores the voices and perspectives of communities that have been historically marginalized and colonized. This exclusion is not only unjust but also counterproductive for the development of an AGI that benefits all humanity, something his book does not seem to dwell on at any point, in a sort of basic USA-centrism.

Returning to de Sousa Santos, his work induces a deep and structured critique of the hierarchies and power structures that have perpetuated the hegemony of Western knowledge. His work is a call for the decolonization of knowledge and the reinvention of power in a way that is inclusive and democratic. When analyzing Aschenbrenner's approach from Santos' perspective, several key points can be highlighted that reveal the problematic nature of Aschenbrenner's approach and the urgent need to reconsider how Artificial General Intelligence (AGI) is conceptualized and managed.

De Sousa Santos argues that dominant knowledge, primarily produced and controlled by Western institutions, has historically been used as a tool for domination and control. This knowledge, which presents itself as universal, in reality reflects the perspectives, interests, and values of Western elites, systematically invisibilizing any other. The decolonization of knowledge implies recognizing and valuing other modes of knowledge that have been marginalized and oppressed by the coloniality of power.

Aschenbrenner, with his focus on national (and, therefore, corporate) security at all costs, perpetuates this coloniality of knowledge by centering on an epistemological framework that prioritizes competition, control, and domination, something characteristic of times in which models like neoliberalism seem to revive in more aggressive versions. His vision of AGI as a tool to maintain the military and economic superiority of the United States reflects a continuity of the colonial logics that Santos criticizes. That approach ignores the voices and knowledge of non-Western communities, thus perpetuating a model of knowledge that excludes and subordinates a large part of the world.

The reinvention of power, according to Santos, involves the democratization of knowledge and the inclusion of diverse perspectives and experiences in the construction of knowledge, of all knowledge, with no single knowledge being key. This would, therefore, imply a break with the hierarchical and exclusionary structures that have historically characterized the production of knowledge and decision-making.

Aschenbrenner's narrative about AGI, with its insistence on security and preparation for future conflicts at all costs, breaks with this democratizing vision. Instead of fostering an inclusive and collaborative dialogue on the development and use of AGI, such an approach reinforces the centralization of power in the hands of a few entities, mainly military and governmental, and less in private ones, a surprising corollary in a researcher of his age who builds his work in a neocapitalist environment, something of which he probably has not been fully aware. The referred concentration of power is precisely what Santos denounces as a form of colonialism that needs to be dismantled to allow for a true democratization and emancipation of knowledge and power.

The coloniality of knowledge, according to Santos text, is not just a historical issue, but a contemporary reality that manifests in how global knowledge is structured and distributed. In the context of AI, this coloniality is reflected in the domination of a few nations and corporations over central concepts for development, such as research or the implementation of advanced technologies. Aschenbrenner, for his part, by emphasizing the need for the United States to maintain its preeminence over other powers, especially China (which he refers to on various occasions), in the development of AGI, reinforces this dynamic of domination. His approach does not consider the global implications of a US-centric AI, which can exacerbate existing technological and economic inequalities between the Global North and the Global South. From Santos' perspective, it is crucial that the development of AGI be based on principles of global justice and solidarity, which include and benefit all communities, especially those that have historically been marginalized.

The United States’ dominance in the development and implementation of AI raises serious concerns for the rest of the planet, especially for Europe and other countries in the American continent. A US-centric AI may reflect and reinforce American values, interests, and policies, which are often at odds with the needs and priorities of other regions, due to the simple learning bias of that AI. From a European perspective, this centralization of AI in the United States can result in a technological dependence that would undermine European sovereignty. Likewise, it can perpetuate global power imbalances and exacerbate existing inequalities. Aschenbrenner's narrative reinforces this trend, suggesting that US national security interests should be prioritized in the development of future AGI, which is problematic for a global community seeking a more equitable and collaborative approach. Europe, for example, may find itself at a disadvantage if it does not develop its own AI capabilities, being forced to adopt technologies and regulatory frameworks imposed by the United States. Should we concentrate on this in the European Union? This would be an interesting element for debate.

An AI dominated by US values and priorities may not align with the general specific needs and contexts of humanity. This is particularly relevant in the realm of security and defense, where US policies and strategies may not be appropriate or desirable for other geopolitical and cultural contexts. Reinterpreting Santos' perspective through this lens, it is fundamental that the development of AGI becomes a truly global effort, which includes and respects the diversity of knowledge and experiences from different regions, cultures, ideologies, or ethnic frameworks. This would imply not only equitable participation in AI research and development but also the creation of regulatory frameworks that reflect the realities and priorities of all global communities, with all that this entails in terms of relinquishing sovereignty for a collective good, something we in the EU know well and value.

FormalPara Curmudgeon Corner

Curmudgeon Corner is a short opinionated column on trends in technology, arts, science and society, commenting on issues of concern to the research community and wider society. Whilst the drive for super-human intelligence promotes potential benefits to wider society, it also raises deep concerns of existential risk, thereby highlighting the need for an ongoing conversation between technology and society. At the core of Curmudgeon concern is the question: What is it to be human in the age of the AI machine? -Editor.