A new Generation of AI Technologies
Developing AI which understands the world.
AI with intuition, common-sense, and understanding context. We are building a new generation AI engine which is completely different from other approaches. In contrast to current methods, our systems are built on the idea that true intelligence comes from language. Learning context using massive raw data and computational power is one way to build AI, but can we do better? Our new form of AI is founded from the ground up using a newly developed concept: everything is language, and intelligence is about understanding the stories arising from that language.
We are not talking about human language, but an entirely new way of viewing the world - as if it were speaking to us. This revolutionary new approach gives significant advantages, including the potential for rapidly understanding novel situations without prior training, detecting anomalies in complex systems and discovering hidden meaning.
Every system has its own characteristics, and yet human experts learn an awareness and understanding of almost any domain, from biological agents, the financial markets, geo-political events, machines and social activities. Our new AI has the potential to understand the characteristics of systems without human intervention. Even when you don't know what to look and aren't able to train your AI system, our models provide a mechanism for capturing awareness.
Emotional AI
Can AI capture emotions beyond biological entities? The human capacity to understand emotion is all about identifying the true message behind the message. And yet the challenge is that emotional features can be subtle, inconsistent, vary across cultures, may be difficult to label and hence present problems for conventional AI. Even if some success can be obtained for human emotions, how can we apply this idea to non-human or other systems? How might we detect when a system is running smoothly or when there are 'choppy waters' ahead? Our novel AI system provides the capability to detect the underlying 'emotion', even for systems which are non-human.
trust
Trusting AI systems is crucial. Our new generarion AI is grounded on recent neurocognitive findings. In contrast to measuring trust as uncertainty, our AI learns what to trust in a similar way to humans. Humans learn to trust through common experiences and stories. This gives an indication of why enabling trusted AI is so challenging. It won't be achieved simply by adding in a module to give some mmeasurement of a 'trust' vector. Instead, it needs to be deeply embedded within the AI model itself. Moreover, it needs to incorporate a form of story-telling capability.
This is where language comes in and the way in which our unique approach offers a powerful mechanism of solving this problem and providing explainability. Full adoption of AI not only requires functional capability, after all, what human is perfect. Rather, we propose that AI which explains what it is doing and builds trust, will gain far greater acceptance with higher levels of safety and longevity.