4 min read
Albeit the older form of AI, Reactive AI is one of the four categories of AI and is often considered the most useful for businesses because it can be configured logically and always produces the same outcomes.
One notable example is Deep Blue, a chess-playing expert system that won against the world champion Garry Kasparov in 1997.
Being the oldest form of AI from the late 90s, it is relative and still far younger than the telephone, which is the most widely used channel today to contact businesses, yet it was invented by Alexander Graham Bell in 1876!
At Spixii, our conversational process automation process is based on expert systems, a proven technology part of reactive AI that is very powerful for businesses following well-defined yet complicated processes to execute.
As an analogy, we consider Expery systems and Reactive AI to be the seat belt of AI, safe, practical, and easily used by everybody.
Like a human expert
According to Java T Point, “An expert system is a computer program designed to solve complex problems and provide decision-making ability like a human expert.”
Expert systems use a knowledge base of facts and rules and an inference engine that applies logic and reasoning to draw conclusions.
One common application of expert systems in decision support is medical diagnosis. These systems assist doctors and nurses in diagnosing diseases, suggesting treatments, and monitoring patients' conditions. For instance, MYCIN is an expert system designed by Stanford University in California to diagnose bacterial infections and recommend antibiotics. It utilizes a set of rules based on symptoms, medical history, and laboratory tests and prompts users with questions to narrow down the possible causes.
The infographic below from INNOVECS highlights its principles.
Only configuration, no training
Unlike generative AI, which belongs to the AI category, Limited memory, expert systems only require the knowledge base of a human expert or customer service expert to be configured. This knowledge is often well documented in documents and spreadsheets used by human experts to solve high-value and complicated processes. Other knowledge might not be documented but is well-known by experts.
At Spixii, for a new conversational self-service, we often start by analysing a spreadsheet, building a prototype and getting feedback from experts who add invaluable improvements derived from their specific experience.
This configuration is the Knowledge base represented by the right rectangle in the infographic above.
Human-friendly interface
Once the system is configured and produces reliable, consistent and approved outcomes, how can users interact with it? In the business world, by users, we mean individuals, policyholders and financial advisors, for example. As expert systems are used to solve high-value processes with a wide variety of combinations digitally, only a human-friendly, intuitive conversational interface can do the job. The aim of the systems is to gather inputs, label them, structure them and process them in order to generate the next action until the queries is solved.
As businesses want to reach their customers 24/7 through multiple devices, computers, tablets and laptops, a conversational interface is the most suited. Indeed, most humans use conversational applications such as WhatsApp multiple times on a daily basis and across devices.
What's next?
At Spixii, our technology doesn't stop at expert systems, as our research & development effort pushes the boundaries of Reactive AI to Limited Memory AI. Indeed, past interaction can be used to refine the experience first, such as displaying specific options that might suit certain demographics better. The aim is to make the conversational self-service even more intuitive and performant.
For more information on conversational process automation and how it can help your customer service operations to save money, you can download a copy of the most recent Spixii white paper ⬇⬇⬇
