Human Mind Simulation Project

We develop human level Artificial Intelligence with the help of volunteers.

.
The project’s immediate goal is to develop an AI system as capable as a 7-year-old child. It understands English and can answer questions in English, just as a child. Our initial goal is a human level Language Understanding system equipped with common sense knowledge.  Read more about the purpose of the project

Why is this different from ChatGPT?

ChatGPT provided proof that human level language understanding is possible. We develop a complementary solution, which focuses on complex step-by-step reasoning, is fully explainable, remembers facts individually and does not hallucinate. These are all missing from ChatGPT. Read more about our analysis on ChatGPT

The HMS Project contributes to AI Safety research by developing explainable AI.

Explainable AI is an important topic within AI Safety research. The ultimate goal of explainable AI is to be able to fully understand the process through which the AI takes an action. With this understanding, we are in a much better position to ensure the AI is aligned and notice if this is not the case, and to intervene, modify the system for better alignment.

In hands-on, practical explainability research, we can take two approaches. One is to attempt to explain the workings of current mainstream AI systems, which means deep neural networks. The other is to develop explainable alternatives to deep neural networks, systems that are inherently more explainable. If we succeed in this, we have a potential augmentation or replacement for neural networks, which may become very significant if neural networks continue to resist full explainability. The Human Mind Simulation Project (HMS) takes the second approach by developing the Cogito natural language understanding system. Cogito is an alternative language model that is designed with a high level of transparency and explainability in mind.  Read more about the Cogito System

We invite students and volunteers to participate – join if you want to make AI safer.

The HMS Project is a large-scale project requiring dozens of participants for multiple years. The project is fully operated by PhD students and volunteers with various backgrounds.  Anyone, from anywhere on the globe is invited to participate via our portal humanmind.uk. Read more about volunteering options