The global conversation surrounding artificial intelligence in healthcare is increasingly focused on a term that evokes both hope and apprehension: superintelligence. This refers to AI capable of surpassing human intelligence in every cognitive task – complex reasoning, data analysis, scenario prediction, and even ethical decision-making. While once the realm of science fiction, achieving superintelligence is now a stated goal for research labs, major tech companies, and global investors. The vision is an AI that can diagnose illnesses more accurately than doctors, optimize entire healthcare systems, predict epidemics, and create highly personalized treatment plans.
Although, even as the world discusses the potential of superintelligence, millions remain excluded from access to even basic AI-powered healthcare tools. The demand isn’t for sophisticated predictive models, but for fundamental resources like chatbots for initial patient assessment, clinical decision support systems, and automated translation to explain medical therapies. The primary barrier isn’t technological, but rather linguistic, cultural, and structural.
Consider sub-Saharan Africa, where over 2,000 languages are spoken, yet the vast majority of AI systems are trained on only a handful of languages – English, French, Chinese, or major European languages. This disparity can lead to patients unable to understand prescriptions, mothers struggling to articulate their children’s symptoms, and preventable medical errors. The situation highlights how AI, intended to reduce health inequities, could instead exacerbate them by overlooking local languages, cultural norms, and communication styles.
One stark example illustrates this challenge: a mother brings her child to a clinic in a displacement camp. She speaks a minority language, while the medical staff only understands the dominant local dialect. Without interpreters or digital tools capable of understanding her language, effective communication becomes impossible, potentially jeopardizing even simple treatments. This creates a paradox – while discussions about superintelligence dominate headlines, basic misunderstandings contribute to errors in rural clinics.
This disconnect was symbolically on display at the India AI Impact Summit 2026, where leaders from major AI companies, including Sam Altman, CEO of OpenAI (the company behind ChatGPT), and Dario Amodei, CEO of Anthropic (founded by former OpenAI members and responsible for the Claude model), shared the stage but did not acknowledge each other. The moment, though seemingly minor, underscored the gap between those developing superintelligent systems and those excluded from the digital revolution – a prioritization of competition and branding over the concrete needs of patients.
In public health, true intelligence isn’t about “knowing everything,” but about understanding. It’s about comprehending minority languages, cultural nuances, metaphors, proverbs, and taboos that describe illness. Without this understanding, algorithms and chatbots develop into impersonal tools, unable to effectively save lives. Errors, poor adherence to treatment plans, and distrust in healthcare services aren’t accidental; they are a direct result of systems built without considering the perspectives of those who will apply them.
Africa provides a compelling case study: a severe shortage of doctors relative to the population, a high prevalence of HIV, malaria, and tuberculosis, and limited infrastructure. An AI capable of understanding all local languages and adapting to cultural contexts isn’t a luxury, but a matter of life and death. However, the development of such a system is hampered by a lack of local datasets, access to data centers, stable connectivity, and adequate training.
The challenge extends beyond language to encompass structural and infrastructural limitations. Africa hosts less than 1% of the world’s data center capacity, and fewer than 5% of African AI researchers have access to the computational systems needed to train complex models or natural language processing tools applicable to local contexts. Without stable infrastructure, reliable electricity, and widespread connectivity, even the most advanced technology risks becoming ineffective. This is compounded by “brain drain,” where doctors, engineers, and data scientists leave local communities, depriving healthcare systems of the expertise needed to build tailored solutions. This creates a vicious cycle: fewer local capabilities, fewer contextualized data, less useful AI, and increased exclusion.
Simply translating words isn’t enough. Health is rooted in stories, metaphors, rituals, and taboos. An algorithm that ignores these elements risks misinterpreting clinical signs, generating false alarms, or providing inappropriate guidance. For AI in healthcare to be truly effective, it must be culturally intelligent, not just computationally powerful.
There are encouraging signs. Initiatives like African Next Voices and Lesan AI demonstrate that investing in local, multilingual datasets yields tangible results: more accurate models, more effective health communication, and improved treatment adherence. However, these remain exceptions. A global commitment combining technological investment, capacity-building policies, and inclusive governance is needed to ensure superintelligence doesn’t remain an abstract concept benefiting only research centers and investors, while those most in need remain unseen.
Before focusing on when superintelligence will arrive, we must ask whether AI will be able to truly listen to all voices. Technological innovation is only meaningful if it reduces inequalities. If it doesn’t, even the most powerful artificial intelligence risks reinforcing latest forms of exclusion. In healthcare, silence is never neutral. Failing to speak a patient’s language means ignoring them, risking errors, and undermining trust and engagement. The real challenge isn’t building machines more intelligent than humans, but creating intelligent systems for all humans, capable of navigating diverse languages, cultures, and contexts. Only then will the promise of superintelligence become ethical, practical, and truly life-saving.
Francesco Branda
Unit of Medical Statistics and Molecular Epidemiology, Campus Bio-Medico University of Rome
