Contextual AI – in search of the holy grail
Over recent years, AI has become part of our everyday lives. We see it in everything from customer service chatbots to the navigation systems in our cars. From these applications, we also see its limitations. The AI bots we use in our day to day lives do not always have a perfect understanding of what we are asking, and this is sometimes demonstrated in their responses.
In relative terms, AI is still new technology. You only need to look at where we are today in comparison to the online help and customer support of a decade ago that was limited to multiple choice questions with two or three multiple possible responses to see how far we have come. Sprout.ai’s Contextual AI will take the technology to the next level, allowing humans and machines to collaborate with better understanding and in whole new ways.
Synthetic data
Part of the difficulty with existing systems is that the AI is drawing its conclusions and making its recommendations on the basis of a finite amount of data. By providing more data, we can create AI that has deeper knowledge and has become more intelligent. But where is all that data to come from? And what are the privacy implications for the owners of that data?
Synthetic data provides an elegant way to provide a limitless amount of realistic data that allows personal privacy to be preserved. It is based on real-world data and so retains its statistical integrity.
Contextualisation
Contextualisation is a whole new way of teaching AI to understand what is put in front of it from a “human” perspective and to consider both the information and the context in which it is presented. It is the combination of understanding a situation and basing it on facts. So here, it means the system ingests data and can use it to understand and explain where you are in a specific process, where you have most likely come from and where things are leading. This is all on the basis of data, but that data can be synthetic.
A new world of possibilities
Harnessing these two exciting areas of development in tandem presents awesome possibilities. It leads us to genuinely intelligent systems that can explain their reasoning processes and be audited accordingly.
It might sound like a concept from the realms of science fiction, but synthetic data and contextualisation mean that our AI can do more than just think – it can imagine. The system can make predictions and explain circumstances without first having to be fed gigabyte after gigabyte of real-world data.
More than that, though, it can contemplate whole new scenarios for which no historical data exists. Society and technology are changing at a faster rate than ever before, and in directions that we can barely glimpse.
Sprout.ai leading the way
So, with these two types of technology, systems can create data from very little, and then explain that data and the conclusions to be drawn from it in a way that mimics human understanding and contextualisation. Sprout.ai is developing proprietary solutions in both areas, and these represent a quantum leap in technical progress.
Our Optical Character Recognition (OCR) solution is recognised as the best in the world right now, even out-performing Google’s API. And it can do it all without first being fed any data at all. This can be contextualised by combining Natural Language Processing (NLP) and knowledge graphs.
Real world applications
We likened these AI developments to science fiction earlier. But the important thing to understand is that today, the technology is real and it has real world applications. These can have a positive impact on every single one of us and our families.
The world of insurance claims is one that we all encounter from time to time and it makes a logical proving ground for this technology. After all, there is wide variation in data, there are as many contexts as there are claims and often there is limited or missing information. We all take out insurance, and after doing so, most of us think little more about it. But when we need make a claim, we are often at our most vulnerable, and the decisions made can have a profound impact on our lives and our finances.
For these reasons, we are convinced that insurance is the ideal place to start the AI paradigm shift and to immediately have a positive impact. But beyond that, the possibilities are limitless and we could soon find ourselves back in science fiction territory. As Paul Brandt famously said: “Don’t tell me the sky’s the limit when there are footprints on the moon.”
