Safety through speed, and the concentration of power
The Logic of the Unstoppable Force
Dario Amodei, CEO of Anthropic, recently asserted that there was no alternative to developing artificial intelligence. This claim frames the advancement of large language models and frontier AI as a natural progression, a tidal wave that cannot be slowed or redirected. By presenting AI development as an inevitability, the discourse shifts from a question of choice to a matter of survival. If the technology is already coming, the argument goes, then the only rational path is to ensure it is developed by those capable of managing it. This logic bypasses the most vital question: why must this specific path be taken, and by whom?
When technological progress is described as an unstoppable force of nature, it functions much like the concept of manifest destiny. It suggests that the direction of human evolution is fixed and that resisting it is not just futile, but irrational. This rhetoric serves a specific purpose. It creates a sense of urgency that discourages slow, deliberate oversight. If a race is already underway, any attempt to pause or implement strict guardrails looks like a delay that allows rivals to win. The \"arms race\" metaphor is not a neutral description of market competition; it is a psychological tool that justifies high-speed deployment.
Historical Echoes of Technological Determinism
History provides many examples of technologies being framed as inevitable to bypass social concern. During the Industrial Revolution, the shift toward mechanized production was presented as an unstoppable economic law. This framing helped consolidate power in the hands of factory owners while downplaying the displacement of workers and the degradation of living conditions. The \"progress\" was inevitable, even if the human cost was high. Similarly, the nuclear age brought a similar sense of inescapable momentum. The development of thermonuclear weapons was driven by a logic of preemptive necessity, creating a global security structure defined by the threat of annihilation.
In both cases, the label of \"inevitability\" helped insulate developers and policymakers from the consequences of their decisions. If a process is inevitable, no one is responsible for its negative externalities. The scars left by the carbon-heavy era of coal and steam are the legacies of a world that believed it had no choice but to burn. Today, the silicon-based equivalent is the rapid, unchecked expansion of massive compute clusters and data scraping operations. By claiming there is no alternative, corporate leaders minimize the possibility of choosing a different, perhaps slower, and more equitable trajectory.
Political Economy and the Concentration of Power
The claim of inevitability obscures the material realities of AI development. Building frontier models requires immense amounts of capital, specialized hardware like GPUs, and vast datasets. This creates a high barrier to entry that naturally favors a handful of massive corporations. When these companies argue that development must continue at breakneck speed to maintain safety or competitive edge, they are essentially arguing for the preservation of their own monopoly. The \"arms race\" is not a race between humanity and machines, but a race between a few concentrated entities for control over the digital infrastructure of the future.
This economic structure creates a feedback loop. Large amounts of capital allow for more compute; more compute allows for better models; better models provide more profit, which leads to more capital. This cycle reinforces the idea that only those with massive resources can safely navigate the AI era. This narrative ignores the potential of decentralized models. If the development of AI were treated as a public good rather than a proprietary race, the incentives would shift from speed and dominance to safety and accessibility. The "no alternative" rhetoric functions to protect these massive capital investments from being disrupted by alternative economic structures.
Sociological Impacts and Cognitive Sovereignty
Beyond economics, there is a deep sociological dimension to this technological rush. As AI systems become integrated into every layer of social interaction, they shape how humans perceive truth, communicate, and make decisions. A rapid deployment strategy prioritizes the speed of integration over the stability of the social fabric. We are seeing the erosion of cognitive sovereignty, where human decision-making is increasingly nudged by algorithmic predictions designed for engagement rather than accuracy or human well-being.
The social cost of this speed includes the rapid spread of disinformation and the degradation of shared reality. If the goal is to win the race, there is little incentive to fix the societal glitches caused by the engine. Instead, these glitches are treated as minor friction in an otherwise necessary movement. This perspective views human society as a passive recipient of technology rather than an active participant in its governance. The dismissal of alternatives assumes that society can adapt to any change, no matter how disruptive, without losing its fundamental cohesion.
The Myth of Safety Through Speed
One of the most common justifications for rapid development is the "safety argument." The idea is that if we do not develop safe AI quickly, a bad actor will develop dangerous AI first. This creates a prisoner's dilemma where every player is incentivized to act recklessly to avoid being left behind. This logic is deeply flawed because it treats safety as a byproduct of speed rather than a prerequisite for it. True safety in any powerful technology comes from rigorous testing, verification, and oversight—processes that inherently require slowing down.
By framing the situation as a race, companies can move the goalposts of what constitutes "safety." Safety becomes a matter of building better filters on a model rather than rethinking the fundamental risks of the model's existence. The rush to deploy ensures that the most dangerous consequences are often only discovered after the technology is already embedded in critical infrastructure. The "no alternative" claim allows developers to avoid the hard work of building systems that are safe by design, opting instead for systems that are merely "safe enough" to pass through regulatory gates.
Alternative Models for Digital Agency
Contrary to the narrative of inevitability, several alternative models for AI development exist. Community-governed AI initiatives seek to place the tools of intelligence in the hands of local groups rather than centralized corporations. These models prioritize the specific needs of a community and the preservation of local knowledge. While they may lack the raw compute power of a trillion-parameter model, they offer a level of accountability and alignment that corporate models cannot match.
Open-source democratization is another pathway. By making model weights and architectures available to the public, the power to shape AI is distributed. This prevents a single entity from acting as the gatekeeper of digital intelligence. Furthermore, human-centric design frameworks suggest that AI should be developed to augment human capability and labor rights, rather than to replace them. These models prioritise the sovereignty of the individual, ensuring that technology serves human ends rather than forcing humans to adapt to the requirements of an algorithm.
Who Decides the Necessity?
The fundamental question remains: who decides that there is no alternative? The decision-making process is currently concentrated in boardrooms and among a small group of engineers. The voices of laborers, marginalized communities, and the public at large are often treated as secondary to the momentum of the market. When we accept the premise of inevitability, we surrender our agency to those who hold the most compute and capital.
To reclaim this agency, the discourse must change. We must move away from the metaphors of warfare and natural disasters. AI is not a storm; it is a tool designed by people with specific interests. We can choose to prioritize different values: equity over speed, stability over disruption, and collective welfare over concentrated profit. Challenging the inevitability of the AI arms race is not about stopping progress, but about choosing a direction that serves all of humanity rather than a privileged few.
