A Nigerian AI Lab?
Why Nigeria needs an AI lab, and why it shouldn't try to replicate what SF and London are already doing.
the argument was so stupid i'm embarrassed i had to have it.
i had an argument recently with a colleague from work in which he told me that nigeria as a country doesn't need an ai lab because of the problems we're already going through - what will an ai lab do for us ?, i feel ashamed even adding this argument to this article because it is fundamentally stupid, an ai lab is needed in fact i believe every country should be studying and exploiting this technology.
but he has a point sorta, but this doesn't mean an lab is not needed, it just means that the problem this lab should be focused on is different, basically we don't need to replicate what SF/London labs are doing. the comparative advantage is in solving problems that matter intensely locally but that wealthy-country labs have little incentive to work on.
we don't need to work on interpretability research, this will be a full on waste of time, we should let sf and london take care of that, as my colleague said the problems we're facing doesn't need attention to that form of research, reverse engineering neural networks to understand their internal computations won't help anyone in this country and so we're gonna move past that.
the comparative advantage is in the problems wealthy-country labs have no incentive to solve.
the first thing i think a lab like this has to work might seem counter intuitive, data infrastructure - but siji no deep learning, neural networks, transformers etc etc shut up, for there to be actual local research we need valuable datasets which we're extremely lacking, there aren't datasets in Nigeria, this brave lab has to step in and see how it can fix that issue e.g , immediately they can begin to generate valuable datasets for what they want to work on and for future local labs to make use of, there is literally nothing more valuable in machine learning that valuable data