Amongst other things, Rainbird is an Cognitive Reasoning-based decision engine and as such can make complex, human-like judgements at scale. You feed it with two things, knowledge and data. Rainbird will then make judgements with or without human intervention, depending on whether it can get the data it needs from other sources. If Rainbird cannot find the data it requires to make a decision, it will ask a human – in plain English.
Under the bonnet, Rainbird’s judgements are based on Cognitive Reasoning which is the basis of most human decision making. While Cognitive Reasoning feels like a human process, it can be expressed mathematically and can therefore be processed by a machine.
What is Cognitive Reasoning?
Here is a simple explanation. I am going to provide you with two pieces of knowledge.
“Socrates is a man.”
“All men are mortal.”
We might contend that the first statement is a piece of data, and the second is piece of knowledge.
If I were to ask you what you know of Socrates, you would tell me that he is both a man and that he is mortal.
But hang on, I didn’t tell you that he is mortal, you inferred that yourself. You joined the dots. That’s what we mean by Cognitive Reasoning.
Rainbird works in the same way. It has a visual interface which enables you to build models of knowledge which it can then infer over, pulling in external data where required. When published, it can hold numerous conversations concurrently, making judgements based on each set of specific circumstances.
Rainbird can also discover new facts, and learn – simply by being used.
It can also express why it has made each and every decision, in plain English. This makes it ideal for making judgements at scale where there is oversight by a regulator, such as in financial services, banking, insurance and healthcare.
You may be reading this blog because you are trying to find an AI solution for your business. Perhaps you are seeking data on which to base a judgement, and that judgement is going to be based not only on the data but your own knowledge – built over many years from your life experiences.
You might take data from this blog, the internet at large, colleagues and other experts – and mash it together with your own knowledge to join the dots and make a judgement. You may ultimately end up with a decision in which you hold a specific level of confidence. If asked you could probably write out your rationale.
This is how Rainbird makes decisions, reporting the outcome with a degree of certainty and a rationale.
Why is this Cognitive Reasoning approach helpful?
This is useful for two reasons:
- Firstly, it is an ideal model for simulating and scaling human decision making because it mirrors how we work.
- Secondly, it is possible for anyone to learn how to model systems in Rainbird, without being a software engineer. It is intuitive because, although we may not immediately recognise it – Cognitive Reasoning is something we all do every day. This means you can build solutions and deploy them rapidly.
Do Cognitive Reasoning-based systems have limits?
Cognitive Reasoning-based systems have been used since the 1980s, but have historically been very difficult to build, requiring software engineers. The maths is very complex and older technologies did not handle uncertainty and ambiguity.
Rainbird model is more powerful and can better represent the real world. It is visual and easy to work with, and you can express uncertainty in both the model of knowledge and the data on which Rainbird is basing each judgement. In fact, Rainbird can make judgements in the face of missing data.
What about Machine Learning as an AI technology?
Machine learning approaches start with a corpus of data, not with a model of human knowledge. Everything has to be discovered. It is ideally suited to finding insights in data that are not obvious to humans. It is an analysis technology, not a decision-making technology.
Can I have my cake and eat it, blending Rainbird with Machine Learning?
Yes you can. Machine learning projects, while non-trivial in size and cost, can deliver powerful insights, but because machine learning cannot provide a rationale, human oversight is still required. Rainbird can be used to apply human-like judgement to machine learning results, giving the best of both worlds.
James Duez is Chairman of Rainbird Technologies.