Rest of World · Technologie

Global AI Experts Challenge Zuckerberg's

Mark Zuckerberg's vision of

Meta has greatly exaggerated the economic benefits of its data centers, both in its recent manifesto and in advertising campaigns targeting small towns. While data centers create jobs during construction, they employ few permanent workers and offer minimal wage benefits to local communities. Furthermore, they can negatively impact health and the environment, issues that are particularly challenging for African countries with fewer protections and less governmental oversight.

Zuckerberg's definition of "everyone" largely excludes marginalized groups, including speakers of minority languages, users of shared devices, and those with unreliable internet access. For AI to truly educate billions, investment must extend beyond large models to include open language infrastructure, consent-driven datasets, community-led language digitization, and compute access for local researchers. Trust in AI will depend on accountable enforcement, independent oversight, and protections that function across diverse legal and social contexts, not just within Silicon Valley's framework.

The promise of "AI for everyone" risks ignoring the reality that not everyone is online, digitally proficient, or able to engage with AI in English. To realize AI's potential in Africa, it's crucial to address last-mile challenges like reliable electricity, high-speed internet, affordable devices, and meaningful access in underserved communities. Investment in African languages and local knowledge systems is also essential to prevent AI from widening digital divides.

Meta's size creates contradictions, especially as the U.S. expects its tech corporations to uphold American hegemony, which can conflict with profit motives. While Meta's document appears to address global regulators with a message of transparency, its latter sections are directed at U.S. power brokers who expect Meta to act as a vector of U.S. power. Meta, as an intelligence company, profits not from increased compute but from using humans to untangle relationships, training a central machine to understand complex webs of relations, which it will use to optimize its machines and serve the U.S. government and its agenda.

Zuckerberg's claim of "AI for everyone" is not new. Decades ago, Facebook promised to connect rural women across Africa, but instead amassed data, monopolized the digital sphere, and controlled the information ecosystem without observable benefit to those women. Facebook has exploited labor, extracted resources, and caused harm through disinformation, while its response to criticism has been to displace accountability and gaslight those it claimed to help. Given Meta's history of silencing criticism, retaliating against workers, burying evidence of harm, and launching problematic products, its claims of acting in people's best interests are laughable.

Meaningful participation in AI requires agency, choice, and trust. Companies must earn trust through actions like designing AI with users, being transparent about data usage, and providing recourse when AI errs. Trust is multidimensional and must be considered from the outset of AI development and deployment for these technologies to truly benefit everyone.

AI-samenvatting op basis van de bron.