IBM XPrize - Cognitive Systems
Towards strong AI with sentient, self reflective "conscious" AI
Would you like to join us in our mission to deliver AGI?
We believe that AGI can be built and that sentience, in the form of a self aware / self reflective network, is a key component.
The aim of this project is to demonstrate a basic form of “sentient” AI which we can use as a seed for AGI. Our vision is to accelerate the pace of change.
Our work is based on in house developed concepts for sentient AI based on continuously active generative networks . We know that we are not the only people currently researching this approach to AGI and are excited that if not us, then one of the other groups could get traction with a seed for strong AI within the next 5 years.
In the AI field it is widely recognized that that the current state of AI is weak, the ability to generalise is poor and our ability to make sense of complex data streams is still rudimentary.
When it comes to dealing with complex data the current generation of AI is insufficient and we need to move to Artificial General Intelligence. For us the key differentiators for AGI are its specific ability to simplify complex environments to make them tractable with limited resource and the ability to generalise from information learnt from its environment.
We believe that experiential learning, with a grounding in sensory reality, is the key to developing AGI. We have developed a model that we believe answers the how and the why of AGI and specifically defines the learning advantage of self aware AGI and interestingly the advantage of consciousness.
Our current status is that we have a high level architecture and a novel embodiment of the AI, which we intend to build and test over the course of the competition. Our aim is to demonstrate the basic core of an AGI.
The idea of how to approach sentient, self reflective, "conscious" systems was a 3 am epiphany which gave us the idea of how the high level system could work and, as a side product, how the experience of “consciousness” probably has a depressingly simple explanation.
Artificial “Consciousness” is not our main project goal, but our model may help to demystify the human view of "consciousness". We are aware of the considerable challenge in ever proving anything in this field, as highlighted by John Seale and recently by Edward Witten. However as AC is just a side product of our work we feel it’s worth including.
The current project timeline is to build out the basic system and the test plan in the first year and refine / recalibrate the project over the following years.
A key challenge will be developing suitable tests to demonstrate the principles and finding a way to do those tests ethically.
Hop over to the IBM X-prize Signup page to join up