One underserved aspect of this debate is bias within large language models (LLMs) behind current AI/ML implementations. 

Africa, for example, is very poorly represented in these data sets. Black skin is poorly served by Wearables and their ability to determine health characteristics for their wearers, reducing health outcomes rather than improving them. 

African languages make up a tiny percentage of language data, despite a billion humans represented within these language groups.  This technology offers the best outcomes for those in limited literacy areas, yet is focused on the markets with the least need for this.

African faces are under represented in CV data sets, increasing fraud and preventing the effective use of biometrics in developing markets - where their benefits could be greatest. 

There is substantial evidence that the prevalence of Sickle Cell characteristics in SubSaharan Africa is a direct and evolutionary response to Malaria's prevalence across the continent. https://www.pnas.org/doi/10.1073/pnas.1804388115

Those with Sickle Cell traits have been able to avoid death from malaria due to the reduced oxygenation of red blood cells due to this genetic mutation. 

While intervention in Sickle Cell Anemic patients is important, attempts to "cure" sickle cell characteristics could have the unexpected consequence of increasing deaths from Malaria by "curing" this evolutionary benefit of this characteristic in Malarian areas.