Dynamic Integrations delivers software integrated with the first steps towards Artificial Intelligence. Implementing this software “Smart Dialog” in a service management environment has led to several insights explained in this blog. It describes the “behaviour” of Artificial Intelligence (AI) in a human dominated service management environment. The type of AI we use is explained, thereafter the challenges implementing it.
Is there a difference between AI and human intelligence?
No, actually there is not! At least not in the way it works. According to the definition of AI the term "artificial intelligence" is applied when a machine mimics "cognitive" functions that humans associate with other human minds, such as "learning" and "problem solving”.
To understand AI, we need to differentiate between intelligence and memory. Everyone is born with intelligence and a memory is built up over time. The intelligence has built Artificial Neural Network (ANN) patterns to search the memory continuously. AI’s huge advantage over humans, is that it has an enormous learning ability (mental power) that will never be met by living creatures. Source Wikipedia
Artificial Intelligence background
Is there one type of Artificial Intelligence?
No, there are many different types of Artificial intelligence including; voice, data, pattern- and picture recognition, which all have different maturity phases and can be seen from different perspectives.
AI used by Dynamic Integrations
The AI type we use at Dynamic Integrations is called Narrow AI due to the limited number of tasks it executes. The memory ANN is connected to, is a database filled with solutions written by humans.
AI in a service environment
The same way humans communicate, AI depends also on good input or a clear problem description. Based on a problem description and specific words used, AI searches its memory (the database) for problems with a similar description and proposes that solution to the customer.
AI (Machine) language
AI “speaks and reads” in its own way. Creating a memory in a language that ANN understands is vital for its performance. Without specific words that ANN can relate to, she will not be able to answer a customer query.
It is therefore of utmost importance that humans and AI “speak” the same language.
Advantage using first line Artificial Intelligence
What makes ANN as a first line agent much more efficient than humans?
ANN’s ability to scan thousands of records in the blink of an eye is unparalleled by humans. No other system can answer any question as quickly as she will, 24 hours a day, every day.
This saves time of service centre agents not having to answer the same questions again. The smarter ANN gets, the more time and money she will save the company.
The experiences using ANN integrated with SmartDialog
Start of using ANN
1. Challenge
The solutions in the database written by service centre agents might contain irrelevant text or spelling mistakes, or not presenting a solution at all for the customer.
AI uses the existing database different because it sends the written solution in the database directly to the customer. Often agents write the solution as a result of a conversation with the customer.
Solution
An agent reviews the database for the most common and repetitive customer queries in the ANN language, and writes a complete solution ANN can send to the customer.
Maintenance of ANN
2. Challenge
New solutions are continuously filling the database, written by agents.
This might confuse ANN.
Solution
Dynamic Integrations has a technical solution for this challenge.
3. Challenge
A changed or updated service might make old solutions obsolete and require new solutions.
Solution
An agent write a specific solution for ANN in the database.
This must be a task executed when an existing service is changed or updated.
Obsolete products/services
4. Challenge
Removal of existing products/services with solutions in the database (memory)
Solution
Question. Do the solutions of the obsolete service confuse ANN, based on the queries send by customers, and there for the answers she gives?
It is always saver to remove the obsolete solutions in the database.
New products/services
5. Challenge
Implementation of a new service.
Solution
An agent must plan and write the most common queries and their solutions in the database for the new product/service.
ANN’s performance must be monitored closely in the start-up phase. The customer queries must correspond with the written solutions to prevent a drop in the customer service experience.
Conclusion
Use of ANN requires understanding of its functioning and new tasks to be implemented in a service centre. These new tasks are primarily focused on maintenance of ANN’s memory, the database.
Next article
This collaboration between agents and AI requires a role within you your company helping AI “to learn”. In our next article we write about new emerging roles in the digital era.
With digital greetings,
Bård Øvrebø, CTO Dynamic Integrations
Fredrik the Frisian, CEO Dynamic Integrations
Comments