An automaton answers the customer service calls to companies, and a bot catches the chat messages. Is this how corporations use AI? It doesn’t need to be so, at least not if you ask the experts at FCAI.
“We want to make Finnish companies understand how pervasive the change AI brings is. It doesn’t mean deploying one piece of AI, but a whole new operational philosophy. AI must be imported to the core of the internal development of the company“, says Arto Klami, assistant professor, from the University of Helsinki and FCAI.
This is how Netflix, the streaming video service, works, for example. It uses AI methods to deduce things from its user data; things like what kind of superhero to have in their next TV series.
This means that the AI software digs through the customers’ watching and predicts their needs, which controls the company’s investment decisions. Corporate directors may use AI to make decisions based on data, instead of having AI only take care of some individual tasks.
Aiming for industries
FCAI recently received a million euros of funding from the centennial foundation of Technology Industries of Finland and Jane and Aatos Erkko’s foundation.
With this funding, FCAI will build AI software that companies can incorporate into the processes they already have in place. The goal is to make this possible without the employees of the company having to understand AI in depth.
The three-year project Interactive Artificial Intelligence for Driving R&D starting at the beginning of 2019 will try to help especially industrial companies to deploy AI.
According to Klami, industrial companies could use AI methods for chemical or other industrial processes, for example, or to simulate the behaviour of people. Many companies are already using advanced models and simulators, but combining them with AI has been difficult so far.
“These past years, we have developed new, revolutionary AI methods and open source software to this end, but we still need further research“, says Klami.
A dedicated professor
The objective of FCAI is not just to make the corporate world understand the possibilities of AI, but also bring the companies into the research as active partners.
“Then we can strive for more considerable breakthroughs, which we will need in the international competition. Our mission is to carry out high-end research into AI in collaboration with the corporate world. We need to produce tools as a result of our research, which will be easy for the companies to use“, Klami says.
FCAI is already collaborating with two dozen Finnish companies. The research centre has appointed a professor dedicated to each partner company to map the companies’ AI needs and opportunities for collaboration.
Unlike Google and Facebook, the customer numbers of Finnish companies are not in the billions, so the companies have to make do with less customer data in their AI experiments. However, according to Klami, we have the skills in Finland to make reliable conclusions from even small data.
“We also have many professors and people with an education in this field, in relation to the general population“, Klami says.
These skills now have to be incorporated in the industries.
“In a small country, it is easy to reach the core of the business, even in large companies. For companies to succeed in the long run, their business has to be guided by a will to deploy AI and the skill to use it.“
This is how AI is changing customer service
AI is changing the nature of many tasks, but not so that we don’t need people any more, says Arto Klami, assistant professor in computer science at the University of Helsinki and FCAI (Finnish Center for Artificial Intelligence).
One of the changes to traditional customer service is having an automated voice answering the service number. It belongs to an AI software based on voice recognition, to which customers may state their business.
There is still a person as background support. If the customer is not able to finish his or her business with the AI algorithm for some reason, the call will revert to a human. It is the human’s task to teach the algorithm that it has failed because the algorithm didn’t recognise a dialectal word, for example. After that, the algorithm will work a little more exactly.
“The human’s work will still be there, though it has changed a bit. This kind of work does not require a higher education than the job required before, as long as the employee has a user interface suitable for teaching the algorithm. There are already companies in Finland offering this service“, says Klami.