I've been reading philosophy since I was a child, and I've always imagined how awe-inspiring it would be to converse with philosophers from the past.

Well, advances in deep learning and natural language processing have made this possible in some ways, and I've set a goal for myself to create a small project as proof of concept.

This project is titled "Speaking with Plato - A Deep Learning Approach to Philosophy." Plato is a favorite philosopher of mine, and his philosophy is still very relevant today.

Plato's Theory of Forms can be seen in the field of pattern recognition. Here we see issues when it comes to training AI algorithms that are easy for humans.

When it comes to image pattern recognition, for example, we can easily train a child to recognize a tree. We can also train an AI to perform this task, but it will fail when presented with a fake tree.

The above problem that AI revealed to us was masked by our meaning-making capabilities. Mainly, our embodied brain, which is dynamically coupled with the environment, can render things obvious to us. Therefore, we think that the explanation comes from that obvious.

Two deep-learning models are used in the project. One is a Chatbot that simulates a conversation with Socrates, while the other is more creative and generates text in an attempt to imitate Plato. All of his work is also explored as part of an EDA (Exploratory Data Analysis).

Here's a sneak peek:

User: What is virtue?

Socrates: A thing which is taught by a certain master, and which is rightly taught by him; and he who taught it, and has taught it also, is good in so far as it is taught?

More can be read in this blog: https://dataspiral.blog/speaking-with-plato/