Home

Origins

Early in 2015 I sent out a mail to a friend of mine debating the state of AI and machine learning in the IT industry. In the mail I lamented that the state of AI in general had not advanced in the past 50 years. As I talked with other friends and associates, I realized that they shared similar opinions. So I began to ask - Why?

The mail got me thinking and has since been the catalyst for the formation of a virtual group trying to figure out how AI could move to the next level and beyond. And although I have not articulated my vision into a condensed vision statement, the mail outlines clearly where I want to go with this.

So.. that is how we got started.

We started with the first question: What limits AI? Is it the computers, or is it the philosophical foundation?

To answer that question, we've taken a whirlwind tour of everything from computer architecture, neurosciences, symbolic logic, propositional logic and epistemology. And because we could not find concrete answers there, we've tried to trace our way back to the roots in Plato and Aristotle. We then traced our path though Locke, Descartes, Kant, Russell and Searle. We were hoping to find answers or pointers to the questions we posed.

Although, we have found no conclusive answers, we've discovered a few important ideas:

• In Aristotle's Categories we find that knowledge also encompasses emotions. Yet, in our passage through Enlightnment to Modern Philosophy, we see a marked shift toward pure logic while eschewing empathy/emotions. Consequently propositional logic has no 'natural' place for them. Therefore, there are no normative representational forms nor are there normative constructs to operate on them. And that directly reflects on modern computer science and computer architecture.
• Kantian philosophy stresses the fact that the mind is intuitional. Yet, the most advanced computing technology is either procedural or at best stochastic. Clearly, a future of emphatic machines imply a fundamental rethinking of how computers should operate today.

Further, we concluded, in the interim, that:

• 'strong AI' is clearly unnatural for modern computers. There are limits. We may be able to work around that limitation and we probably will in the immediate future.
• Empathic machines require a fundamentally new systems architecture for them to be intuitional.

This project is an attempt to discover how we could remedy that.

Objectives

Project Kant is about implementing a universal inference engine, based on Kant's first Critique; a so called 'Kantian' machine. This is to better understand Kant (he is hard to comprehend and what is more, he leaves a lot of room for interpretation) and to verify that Kant's claim that his system is complete
is indeed true.

Influences

There are several influences on this project. We draw ideas from computing science, philosophy, mathematics, and engineering. Some of the key pieces of work, publications, or ideas that have influenced this project is categorized in the references page.

Notes

The notes page contains a collection of design notes, observations and design challenges that are relevant to this project.

Status

The project is currently in early planning stage.

Dependency

The Kant Project is built using the Empact Foundation Class Library.

Project Members:

• Sreekant Sreedharan (admin)