Mind the AI Gap: Understanding vs. Knowing

Next week I will be speaking at the 2024 International Business Pedagogy Workshop on the use of AI in education. This gave me the opportunity to put together some ideas about using LLMs as learning tools.

Some key points:
– Knowing is not the same as understanding, but can easily be confused.
– Understanding requires the ability to reason and identify causality (using counterfactual thinking).
– This is completely lacking in LLMs at the moment.
– Framing LLMs as magical thinking machines or creative minds is not helpful because it can easily mislead us to lend our cognition.
– The best way to benefit from LLMs is to recognize them for what they are: Master memorization models.
– Their power lies in their sheer processing power and capacity, which can make them excellent learning companions when used properly.
– How can LLMs best be used as learning companions? That’ll be part of our discussion.

Source