AI of the Oppressed
One of the most common uses for Generative AI for me personally is as an instruction manual, or maybe a teacher who doesn't care if I interrupt and won't judge me for asking too many questions. Teachers never seem to mind but I always feel a little bad about it anyway. AI is not a teacher, to be clear. If anything it’s more like a textbook: it provides access to information, it provides a new perspective you hadn’t thought of yourself. But at the end of the day you still need to pick up the book, ask the question, turn the page, read the words. And so, in the vein of education I now introduce you to Paolo Freire, adult education theorist.I suggest that his ideas of the contrasting “banking” and “conscientização” modes of education have a place in any discussion on AI as an educational tool. You have a choice whether to use AI under the banking or under the conscientização framework.
The banking mode of education refers to the practice of teaching information (deposit) and then reciting (withdrawal) of that information later. Standardized testing looks like this, but any kind of rote teaching qualifies. In contrast conscientização is roughly education that leads to the learner gaining an ability to produce and affect their reality. The major risk here is that in using AI you may undermine your own agency.
Some people seem to use AI merely to get the answers or do the work. They get from A to B as fast as possible and don’t feel a need to understand anything along the way. The biggest symptom of this isn’t the topics they ask about, it’s the mental model they bring. I believe those who think they have a partner in AI, or an assistant or an intern, or even a friend are offloading their agency. Because they now feel like there’s a helper they are willing to hand off responsibilities to that helper. The helper, the AI, is a multibillion dollar product designed and built by a massive corporation. It’s trained on the collected digitized works of mankind. It is therefore influenced by history, it is influenced by corporate policy. And unlike your grade 3 teacher or your university professor it has no idea what you want or how to help you. It has no ideas at all. It can provide answers to questions, and it can write a chapter of a book for you, but you’re a human and you need to practice to improve. Without practice your skills will fade. More than that, it can never be better than you at being you. Each person is a collection of experiences and skills, and no two are identical. Your voice comes from that. So your voice is unique in all of history. AI can never be your voice. Even if you use it to write you still need to be there, in the shadows, pressing buttons. It can do something that looks nearly identical to speaking in your voice, but it’s a simulation, not based in meaning. So what’s the problem? By using AI to write, you don’t express yourself fully. By trusting AI’s “opinion” you stop making decisions yourself. And if your only source is AI you limit all your informational input to a single channel.
So what’s the other option? Conscientização is learning that expands the worker’s control over producing and adjusting the world in which they live. In this case you can still use AI, but you use it to build your capabilities. Ask it questions, read the answers, decide if you agree or not. Challenge the questions. Challenge the premises of the questions. You must understand the tool for this to work. Otherwise you’re leaning on a magic crutch hoping the pixie dust doesn’t wear off. Challenge the tool itself. Challenging the tool could take many shapes, but it includes being willing to believe the AI could be giving you incorrect answers, and understanding how the system works so that you know how to interpret the answers.
And how does it work? AI is just a computer program. Most programs are deterministic (Do X when the user clicks the mouse), this one is similar but with a bunch of probablistic elements thrown in. It does not, and from what i can see never will, replace your judgement.