UNO November 2019

The revolution of chatbots and artificial intelligence

The 2019 report of Chatbots Magazine describes some of the most well-known trends on the chatbots market in the USA, which I now summarize. Business Insider predicts that by 2020, 80 % of companies will be using chatbots. Lauren Foye affirms that by 2022, banks will be able to automate up to 90 % of their interactions with customers through chatbots. And Spiceworks assures that 40 % of the large enterprises that employ over 500 people plan to implement one or several smart assistants based on Artificial Intelligence in corporate mobile devices during this year, 2019.

All companies are forced to introduce AI in one form or another. It is the electricityof the future. Chatbots are one of the developments of Artificial Intelligence that will most clearly help to lead many key facets of the company, such as customer recruitment, care, interaction or engagement; sales and eCommerce; internal communication; and, in general, the automation of numerous processes with major cost savings.

It is nevertheless true that malpractice can adversely affect the enormous potential that chatbots can bring companies and brands throughout the world. I will highlight the qualities that a chatbot should have:

“Chatbots are one of the developments of Artificial Intelligence that will most clearly help to lead many key facets of the company”

1. A conversational chatbot must be based on, but not conditioned by, Artificial Intelligence. Some of the existing chatbots are directed, not conversational, with negligible potential. Other AI  based experiences, like Microsoft’s Tay, had to be withdrawn just a few hours after launching due to offensive bias in responses. The use of AI in chatbots must be controlled, efficient and subject to the company’s goals. Bearing in mind the limitations of Natural Language Processing (NLP) in AI at present, each chatbot requires an enormous professional effort in its technological and methodological aspects.

2. Quality, appropriate responses. One of the aspects that users find most frustrating is the high percentage of wrong answers. A chatbot correctly designed and trained must correctly answer at least 90 % of the questions asked.Very few technology enterprises can commit to delivering this goal at this time. The quality of the responses assessed by specialists and confirmed by users is key to a successful chatbot.

3. Educate effectively and in record time. Although hundreds and even thousands of theoretical questions are taken into account when designing a chatbot, positive results that are significant for the company will be reliant on training with real users. Technology and ad hoc tools are essential to achieve that effectiveness in record time.

4. Chatbots must perform useful actions. Apart from informing, chatbots must help users to perform tasks in a simpler, easier manner. With voice or in writing. Actions such as subscribing, booking or buying in a restaurant, hotel, plane, shop, clinic… Or effecting transactions at a bank or public administration. Therefore, it must be possible to integrate them into their computer, CRM, etc. systems and perform all kinds of operations securely.

5. Visibility and integration. A chatbot must be available 24 hours a day, 365 days a year with the utmost visibility. Its integration in all channels with an optimized design is essential. It must be efficiently integrated in the web system, apps, social networks, communication tools (WhatsApp, Slack, Telegram, etc.), home, wearables… Low levels of interaction may be due to deficient integration.

6. Data and personalization. A specific user’s data must serve to improve the service, providing users with an increasingly customized product, tailored to their personal preferences, likes and needs. This is the main comparative advantage. An investment in chatbots without this goal is largely a squandered investment. Artificial Intelligence will play a decisive role in individualized personalization and engagement based on the big data generated by the users. Advanced techniques are already available with very productive results.

7. A specific chatbot for each use. Its design must be closely linked to its use and goals. For the time being, AI -enabled technology is still just a means. A chatbot must be specifically designed according to its goals or tasks. Four types of chatbots must be designed very differently, depending on whether they are to be used for eCommerce, marketing, customer recruitment, communication or a call center… Each one has a very different mission and must be given very different technological and methodological treatment. Otherwise they will be a waste of money.

8. Measure and analyze. Measurement and analysis are essential. Companies must constantly improve their responses and impacts among customers and employees. Integration between employees with experience in product, marketing, customer care and human resources and AI specialist teams is fundamental. The designing of virtual assistants can be made more effective if this collective, multi-discipline work is assessed with metrics and analyses that clearly identify the solutions and improvements to be made, based on the data provided by users in their interactions with the chatbots.

“The use of AI in chatbots must be controlled, efficient and subject to the company’s goals”

9. Transparency and ethical principles. In order to gain the trust of our customers or employees, we must be transparent and apply sound ethical principles and values. The AI potential must be used and advanced chatbots implemented with a commitment to guaranteeing a totally responsible, ethical use of users’ rights.

Few tools currently on the market are able to meet these nine requirements efficiently (low cost and value generation). Technology needs to be combined with best professionals in product, brand and customers (chatbots for customers) or human resources (internal communication chatbots). But even at the risk of making mistakes, all companies should already be playing this game.

Chatbots are being designed in Spain with unprecedented effectiveness and results. Harvard Business Review currently reveres Spanish technology as a model of success in the sector. With a rate of correct answers of more than 90 % and using the most advanced platforms on the market, they focus on the qualities mentioned above. This is an opportunity that should be taken up by companies in Spain and around the world.

Andrés Pedreño
Professor of applied economics and Chairman of Ityis, 1MillionBot and founder of Torre Juana Open Space Technolog
Professor of applied economics, expert in digital economics, entrepreneur, Rector of the University of Alicante (1993-2000), CEO of Universia (2000-2004), and member of the holding council of Santander Universities and Universia up to the present. He is a member of the government’s group of experts to prepare the white paper on Artificial Intelligence and Big Data. He was a pioneer in the introduction of MOOCs in Spain through the Unimooc project. Chairman of Ityis, 1MillionBot and founder of Torre Juana Open Space Technology. He has written over fifty books and specialized publications on economics. Eisenhower fellow 1988, Dr. Honoris Causa from Nottingham Trend University (1997), award for scientific merit 2016 from the Valencian regional government (Generalitat Valenciana), Academy of Moral and Political Science, North American Academy of the Spanish Language. [Spain] 

We want to collaborate with you

Do you have a challenge?

Would you like to join our team?

Do you want us to speak at your next event?