According to a talk by John Vervaeke, there are four ways of knowing. Of these four, there's only one that Generative AI (GenAI) and LLMs, like ChatGPT and its competitors, can accomplish (sort of).
When we think about educating our young people with knowledge in mind, expanding our notion of what it is to know will open our approaches to teaching and learning in ways that will continue to develop their ability to think both creatively and critically.
For all those involved in schooling who have been obsessed with instructional practices for the past 40+ years, this is a significant point of tension. For those of administrators who are “data-driven” and inadvertently squeezing the life out of classroom teaching, consider the harm you are committing to the well-being of those you lead. Use this scale for ways of knowing as your guide.
Here is an overview of four types of knowledge that can help keep education relevant now and into the future.
1. Propositional Knowing (Knowing That...)
This is what GenAI is really good at mimicking because they have trained on a data-set, a whole internet database of information, testimony, facts, and opinions.
When you ask it something, or prompt it to provide an answer, it's able to come back to you with the information that is available to it, and what you are looking for. These are static descriptions or generally accepted value-judgments about the world. They sit in the form of propositional statements or declarative sentences.
Examples:
Knowing that Ottawa is the capital of Canada.
Knowing that cats and dogs rarely get along.
Knowing that AI is disrupting education
Knowing that squares have 4 equal sides and four ninety degree angles.
Some are true, some are false, some are debatable.
For the debatable sentences, typically GenAI will draw on statistical occurrences to determine its own position on the matter.
GenAI is able to produce an impressive amount of text but typically it only does so in a way that lists, describes, explains, and compares the different information that is already available to it. It is not able to produce novel insights that go beyond the intentions or the goals of the user.
2. Procedural Knowing (Knowing How...)
This is the type of skill building that we do or the skills we have as embodied beings.
Examples:
Knowing how to ride a bike
Knowing how to use language fluently and effortlessly
Knowing how to driving a car
Knowing how to train yourself to react a certain way in any particular sport
Knowing how to follow a routine in synchronized swimming
Knowing how to give a live performance on stage
Knowing how to think and act like a scientist
Certainly, GenAI can provide us tips and manuals on how to accomplish this type of skill building, but it's not something that can just be automatically infused into your body and inform immediate habitual action.
3. Perspectival Knowing (Knowing What It Is Like...)
This is where it gets interesting. The experiences you have that go beyond information processing.
Knowing what it is like to have a sense of self, understanding your situation in the world compared to objects and others around you. What your mood or state of mind is and how that affects your ability to function in the world. It's on the plane of social reality that is inaccessible to any form of AI (currently and probably for a long time).
Examples:
Knowing what it's like to see the color red
Knowing what it's like to taste a strawberry, the sensation of sweetness it provides you
Knowing what it's like when a scent can conjure up an old forgotten memory
Knowing what it is like to be discriminated against
Knowing what it is like to be excluded from your peer group
Perspectival knowing situates us in pondering what is meaningful, what should be noticed, what is prominent in ways that only a person who has a sense of self and a body with nerve endings and agency that allows us to maneuver in the world with other sentient beings in a way that is deeply purposeful.
There's no sense of purpose in propositional knowing. Generative AI or LLMs don't have a sense of experience.
One way we might try to correlate perspectival knowing with propositional knowing is aligning facts about the world with how we experience them. The simplest example I can think of is if you taste a strawberry and it tastes really sweet compared to another strawberry which isn't so sweet. One way you can correlate that is measuring the sugar content in both strawberries and applying a measurable number to it in a way that makes it propositional. The significance of that data from the propositional side is only important to you because of the level of sweetness you experienced and have concerned yourself to investigate further.
4. Participatory Knowing (Knowing by Being...)
Participatory knowing is a blend of perspectival knowing and procedural knowing. It goes back to that idea that we have a sense of self and navigate the world around us in a way that preserves who we are, but also in a way that we feel inspired or in the way that we aspire to be greater than what we are.
Examples:
Knowing what it means to be a parent or teacher
Knowing the importance of nurturing the community you belong to
Knowing how to be a leader, not just act like one
Knowing by being someone who your friends can rely on
You act in the world, you're changed by it, you influence others. There's an impression you can make in life that can resonate throughout the rest of history or sit prominently in the memory of others.
It is a type of knowing that can leave a deep effect on the world in an affectual way.
A Framework for Parents to Ask Questions
Scale for ways of knowing
When thinking about how to raise our children to acquire knowledge in the world, consider these four ways of knowing. Notice how they build on each other.
Treat it as a scale or a taxonomy of forms of knowledge: from ones that are easy to acquire to ones that take a whole lifetime to build.
If you're a parent wondering how your child’s school is responding to the disruption of AI, use this framework to assess how they understand knowledge—and ask how they’re ensuring your child experiences knowledge acquisition that is both procedural, perspectival, participatory.
Reference:
Vervaeke, J. (n.d.). John Vervaeke - The Four Kinds of Knowing [Video]. YouTube.
Thank you for this classification. I can embrace these ideas, and conversations like this: https://open.substack.com/pub/richarddawkins/p/are-you-conscious-a-conversation?r=f8p8m&utm_medium=ios, lead me to wonder whether this classification does really separate our knowing from AI’s “knowing.” Acts of caretaking, physical interaction, movement still seem uniquely human. I also like your distinction about how you derive knowing (experience or math (for a gross simplification)) but then Dawkins’ post makes me question that even philosophically—is knowing derived from statistical prediction not also a way of knowing? Does it really matter how we come about our knowledge?
Nice article! I’ve been a school governor (in the UK) for a number of years and this information would have been a very helpful input into leadership discussion. I am myself very close to these subjects but most people I worked with were not…