A friend of mine recently posted on Facebook, “Why are there leftists who think Google is an appropriate means of anti-capitalist, anti-oppression pedagogy?” The first comment started talking about learning algorithms, as if such algorithms could somehow be politically, perhaps more importantly, contextually neutral.

Part of the discussion included the question, “If an encyclopedia reports reasonably accurately on, for example, slavery, does that make it pro-slavery or anti-slavery” and went on to say, “Google is a source of data, and in some cases, information”.

This of course leads to the question of whether learning algorithms are truly. One person shared a link to the bookWeapons of Math Destruction. I added to the discussion with a link to AI programs exhibit racial and gender biases, research reveals.

It seems to me that learning algorithms reflect the social context in which they were constructed. Unless there is some conscious effort by their creators, they end up re-enforcing the dominant narratives. This is than exacerbated if they factor in the choices of the users emphasizing the filter bubbles we all live in.

I must admit, most of what I know about Foucault comes from Googling him and reading various Wikipedia articles, but it seems like part of the response to this discussion is that people should Google Foucault; maybe even throw in some Deleuze, Guattari, Derrida, Fanon, Lacan, and others.

Then, as we move past learning algorithms, social constructs, and dominant narratives, perhaps we need to Google Freire as well, but that probably deserves its own blog post.