thinking-robot

We have developed an innovative conceptual and computational framework for artificial general intelligence. The model is called RTOP which stands for Reduction of Temporal Observations and Preditions. The details about our approach and philosophy are discussed in our pre-print paper http://arxiv.org/abs/1910.10393

We are actively developing an implementation of the model, which can be trained by interacting with it, the same way a human baby is trained. Eventually it should be able to generalize reasonably well, that is, be able to apply the past learning in new circumstances for optimum returns; and it should be able to generate novel representations and thought. While we have taken multiple steps towards concrete implementation of the same and have made good progress, we are making available a limited feature demo program (GI-Spark) to make it easier for interested parties to understand our ideas and approach. A quick introduction to the program is available here

GI-Spark Introduction Video

We are looking for partners who share our vision and are willing to participate in taking this effort forward. If you are interested, you can write to us at support@akribiatech.com

The demo version of the program can be downloaded by clicking the below link. The program is written in Java and utilizes JRE1.8-32Bit. To run the program, first the zip download needs to be extracted. On a Windows machine, the program can be launched by double-clicking the GI-Spark.exe file. The program requires the machine to have a camera, a microphone and a speaker.

GI-Spark Download