AI's Existential Fears Mask Real-World Harms and Fuel Digital Colonialism

As artificial intelligence rapidly advances, it presents significant societal challenges. Companies are already implementing layoffs as AI models can now perform tasks previously done by humans. The expansion of data centers has led to increased energy consumption and a larger carbon footprint. Malicious actors are utilizing AI agents to automate the dissemination of disinformation, and algorithmic bias raises concerns about flawed applications and discrimination.

Furthermore, AI's use in identifying targets in warfare contributes to greater loss of human life. These pressing issues necessitate serious regulatory attention. However, current dominant AI policy discussions often overlook these immediate challenges, instead being dominated by apocalyptic prophecies. This doomsday narrative is promoted by some within the AI industry and echoed by high-level decision-makers, including Senator Bernie Sanders advocating for superintelligence bans, cautionary messages from the Vatican, and former Secretary of State Hillary Clinton labeling open-source AI a significant threat.

While it is reasonable for global leaders to heed these industry-amplified warnings, the focus on AI's "existential risk" has inadvertently diverted attention from the multiple crises AI has already created, impacting labor, the environment, and communications. More concerningly, this public panic is being exploited to enforce a modern form of digital colonization, concentrating power within a few AI corporations in select countries.

The doomsday narrative serves as a convenient justification for regulations targeting open-source AI models, which are more accessible and adaptable. The argument presented to lawmakers is that if advanced AI poses a potential weapon of mass destruction, its development should be restricted to a select few. Under the guise of "safety," tech giants aim to solidify their monopolies by stifling competition from smaller, more affordable open-source AI models. This strategy mirrors historical colonial practices, where local industries were prevented from developing, forcing reliance on the colonizer's products.

Currently, the entire world, particularly the Global South, which lacks the resources to develop its own AI models (costing upwards of $100 million in compute power alone), is dependent on these frontier AI systems developed by tech giants. The issue with these dominant models extends beyond their cost; they often perform poorly in local contexts. For instance, Kenya's recent rollout of the AI-driven Social Health Authority (SHA) system failed due to an imported model based on flawed assumptions about Kenya's informal economy, leading to systematic overestimation of incomes and overcharging of the poorest citizens for healthcare.

For frontier AI systems to be effective, they must be adapted to local data. Since building them from scratch is financially prohibitive for most, open-source AI offers the only viable pathway for local innovators. Crucially, preserving open-source AI empowers the Global South to move beyond the West's focus on human-mimicking chatbots. Developing nations require AI solutions for challenges in energy, healthcare, agriculture, manufacturing, and climate mitigation, not necessarily conversational AI. Suppressing open-source AI out of an exaggerated fear of AI agents deprives developing nations of a key mechanism for building their own sovereign tech ecosystems.

Moreover, AI monopolies create a cycle of data extraction and severe cognitive dependency for users. We are already observing early signs of cognitive degradation as individuals increasingly delegate their critical thinking, creativity, and agency to frontier AI systems. This dependency fuels further data extraction, leading to more capable systems that erode human agency and centralize control.

While the Global South is particularly vulnerable due to historical inequities, this dependency poses a threat to everyone not at the technological frontier, including wealthy nations in the Global North. Calls for AI bans are not the solution. Abandoning AI would mean forfeiting a technology with immense potential for accelerating drug discovery and addressing energy and climate crises, as evidenced by tools like AlphaFold and MOFSeq. Banning open-source AI models is equally detrimental. Safety should not come at the expense of equitable access to AI. We must invest in robust safety research and rigorous testing for near-term vulnerabilities. However, we cannot permit fear of hypothetical futures to justify the present consolidation of power.

Safeguarding AI's future requires decentralized, transparent, and globally distributed development. It means championing open-source innovation as a public good and a vital defense against digital colonization. It involves proactively addressing the cognitive impact of AI, ensuring it augments human intelligence rather than diminishing it. Ultimately, it means establishing clear human accountability, rather than attributing agency to AI systems devoid of humanity.