A Reversed Paradigm: Law as Infrastructure
Nairobi has recently removed a cornerstone of digital control. The Kenyan Court of Appeal declared unconstitutional the criminal penalty for spreading false information online, a move that dismantles a mechanism used since 2018 to stifle bloggers and journalists. This decision is not only a victory for content creators but also a formal recognition that truth cannot be regulated as a consumer product. The paradox arises when considering that the same month, a fraud syndicate launched 160,000 identity theft attacks on African fintech platforms, leveraging AI to generate malware at zero cost. The vulnerability lies not in the system itself, but in its relationship with the human context.
The Kenyan case reveals a structural contradiction: while institutions seek to contain AI with legal tools, criminals use it to bypass the same rules. This duality also manifests in the financial sector, where African startups spend years preventing the opening of fake accounts, only to discover that the main threat is internal to the system. The relationship between technology and regulation is not linear but a field of forces where every advancement generates new vulnerabilities.
Architecture and Vulnerability: The OpenAI Case
GPT-4, the most advanced model from OpenAI, introduces a control paradigm that seems to contradict its very nature. The ability to handle 1 million tokens and support advanced coding suggests an extremely robust structure. However, the same model has revealed limitations in managing complex decision-making processes, with studies showing how reasoning chains are difficult to control. This dual aspect – power and fragility – is also reflected in the business context, where the partnership with the Department of Defense triggered a user exodus of 2.5 million, highlighting how trust is not built solely on technical performance.
The scalability of GPT-4, combined with its ability to integrate with tools like Excel, suggests a trend towards the ubiquity of AI. However, this diffusion is not neutral: each implementation introduces new points of failure. The case of Ray-Ban Meta Smart Glasses, where the use of recorded images to train the AI raised issues of consent and data protection, demonstrates how technology does not exist in a vacuum but interacts with complex social and legal contexts.
Human Voice and Machine: The Ongoing Debate
“Advanced economies expand and remain competitive not through additional labor inputs but through capital deepening, technological progress, and total factor productivity growth.”
Jun Du, economic strategy
The words of Jun Du highlight a vision in which innovation is not an abstraction but a mechanism for transforming work. This approach clashes with the reality of African startups, where AI does not replace work but reconfigures it. The case of Wilbe, which provides physical infrastructure for scientific startups, shows that innovation requires not only algorithms but also material resources. The distinction between human capital and technical capital dissolves when we observe that both are necessary for sustainability.
“By adding physical lab infrastructure, Wilbe can now support scientists end-to-end.”
Wilbe, support for startups
The statement from Wilbe reveals a strategy of integration that goes beyond pure AI. Access to physical laboratories and targeted funding allows researchers to overcome logistical and technical barriers. This model, applicable to the Kenyan context as well, suggests that resilience does not stem solely from advanced technology but from a combination of material and intellectual resources. The challenge is no longer to choose between humans and machines but to design systems in which both collaborate.
Scenario in 3-5 Years: The Invisible Map
The year 2026 marks a turning point in the relationship between technology and governance. The removal of repressive laws in Kenya and the expansion of AI in critical sectors suggest a trend towards normalization. However, this normalization is not a linear process. Every technological advancement generates new ethical and practical questions. My impression is that the future will not be defined by individual tools but by the ability to build infrastructures that integrate technology, regulation, and social context. Resilience lies not in predicting the future but in designing systems that can adapt to it.
The map that emerges is not a utopian vision but a field of tensions. On one hand, AI offers tools to solve complex problems; on the other hand, it introduces vulnerabilities that require new forms of control. The true paradigm shift is not technological but epistemological: the recognition that technology does not exist in itself but in relation to human contexts. This is the ground on which the next decade will be built.
Photo by Google DeepMind on Unsplash
Texts are autonomously processed by Artificial Intelligence models