Arseny Kondratyev — head Liquid Studio Accenture Russia
The concept of “chat-bot” came into widespread usage in the early 1990s. However, the concept and technologies for its implementation date back to the 1950s, when such solutions wrote Alan Turing and Joseph Weisenbaum. Today we see that the chat bots artificial intelligence is developing rapidly. Moreover, this area may soon be a breakthrough. In the most General terms, a chatbot is a computer program with which you can communicate through the interface in the form of a messenger app, a window for chat or using your own voice.
The first chatbot was developed in 1966 and was called ELIZA, was a simulator therapist, answers to common questions that users have loaded into the machine in a printed form. Over the years this principle of communication has given a new shape, improved and eventually evolved into modern formats.
The contemporary surge of interest in the chat-bots is due to two key factors: the development of messaging apps — Facebook Messenger, WhatsApp, WeChat, LINE, Telegram, and successes of applied research in the field of artificial intelligence (AI). Most large companies have invested in the development of platforms for chat-bots a substantial amount, you can select Apple, Google, Samsung, Microsoft, Amazon and Facebook.
According to a study by Accenture, CIO worldwide believe that the chat bots will play a critical role in the IT— and business architecture of the company of the future — 56% of respondents said that the quality of the chat-bot can seriously “pump” business processes in their companies. 57% also believe that a well-implemented project functional chat bot capable of providing a significant return on investment with the minimum of effort.
Modern chat-bots has taken a big step in the development compared to the samples from the relatively recent past. If the first bots were built according to the algorithm “simple question from a limited list — a simple answer”, but today they can be programmed to perform very specific tasks, improve operations and reduce their cost. If to speak about advanced bots, they, unlike the counterparts of the previous generation, can learn and reprogram itself for the new tasks. While this is only the first generation for AI platforms, but even they have the ability to perform complex tasks.
Within 2019, the analysts admit the high probability of appearance of the first bots that work without human intervention, solving complex problems through decision-making relevant to the context. The next development step will be the ability of one chat-bot to seamlessly navigate between multiple AI-interfaces as a fully-fledged cross-platform tool. These bots will also be able to remember the context of previous sessions of communication with users to generate answers on this basis.
From chat bots in virtual assistants
Thus, the main vector of development of the industry today — a request for a chat-bots with whom communication is not implemented on the principle of “command-bot”, and through normal human speech. That is, creating chat-bots, modern developers should strive not just to provide functionality at the request of the customer, but also to bring the communication with the bot to the experience of talking with a real person. So today in practice, many companies are increasingly the term “virtual assistants”, while the “chat-bot” is gradually giving way — that the displacement of the “bot”part of these decisions is one of the urgent tasks of direction.
Image recognition, voice, natural language processing using AI — all of these areas demonstrate today high precision: the percentage errors in the solution of tasks on the recognition of the means of AI at the annual ImageNet competition has fallen over the past six years from 25% to 2%, which is higher than in humans.
These possibilities are what create the conditions for a transition to a new level: in Accenture Liquid Studio designed the guidelines for creating voice and text virtual assistant, principles of interaction which are based on scientific research and experiments.
For example, today it is especially important to understand that the style of voice and text communication are very different, it is impossible to cover one functionality, so even when using a single channel, such as mobile apps, a virtual assistant needs to understand how to apply it, and accordingly to build a dialogue and replicas.
Clever bot = satisfied customer
As soon as people understand that the functionality of the bot has reached the limit and he is unable to help them further solve their problems, followed by the completion of communication. The client expected the solution to the problem by high technology, but in the end just wasted time.
A typical scenario: you created a bot for the commercial Bank, answers questions about the transaction account of the client. On the one hand — cool: the algorithm works reliably and gives clear answers to queries such as “How much I spent on food last trip to Europe?”. On the other — as soon as the client clarifying questions “what is this cancellation from April 10 at 3 Grand? I don’t remember!” or “Why didn’t anyone warn me about the fee when I opened the account?” — system stops working. The bot usually does not understand the questions, do not perceive them as a treatment to yourself.
Any clarification can easily enter a chat bot into a stupor, and in such situations the need for its “humanization” is felt in full.
However, changes in the direction of improve digital assistants come, being realized through the refinement of existing algorithms of recognition of natural languages and the development of new, such as Conversational AI. This technology of artificial intelligence, specifically focused on a variety of tasks from precise identification of issues in relation to the activities of the company prior to the step of selecting correct answers on the most difficult cases until the changes to the terms of the mortgage in the Bank. Conversational AI is developing rapidly and becoming more complex, which allows much more effectively to ensure productive relationships with partners and customers business in an informal, friendly manner.
Business benefit expectations and realities
Despite all the challenges facing today’s direction, digital assistants have bright future the potential benefits of their use is too obvious to be ignored. The main reference here can serve as a study eMarketer: number of Millennials in the U.S. make use of every day at least one voice assistant, from 29.9 million in 2017 will reach 39.5 million in 2019.
Speaking about specific results, there is an example of Amtrak, which through the use of chat bot for managing interactions with customers saves $1 million a year for the expense of email marketing. Compared to the cost of development and support chat bot is eightfold gain in money. The company Charter has reduced the cost of customer service due to the bot by 44% in the first year of operation and achieved a 5x ROI in the first 6 months of operation. Knorr reported a 50% increase in sales of its soup cubes in 3 months after starting the bot, advising homemakers for recipes using their products. 91% of companies are either already using or developing its attendant, expect a return on investment in its creation up to 5 times in the first 12 months of operation.
6 out of 10 respondents Accenture executives believe that virtual assistants can improve the ability of their company to efficiently handle incoming queries by interacting with other bots (e.g., from other departments and partner companies), through higher level of attention to visitors of corporate websites at the expense of better dialog options, etc.
Over 50% also believe that bots will help improve the quality to attract new customers and increase loyalty of existing ones. 61% believe that the chat-bots and voice assistants increase the productivity of the employees, 57% are confident in their ability to improve the quality of internal communications. By the way, a study of the attitudes of top managers about chat-bots, shows no trend in the substitution of people — managers are least likely to have indicated a similar scenario for the use of chat-bots in their practice (10%).
Most often today, virtual assistants are adopting the company of the spheres: healthcare, Telecom and Finance. In most cases there implemented chat bots informants. About 25% use or develop assistants aimed at improving the efficiency of the company, as well as bots, are able to carry out the transaction or the sale without human intervention. Regardless of the type of embedded assistant the overwhelming number of them focuses on client interaction, followed by assistants with an emphasis on after-sales product maintenance, CRM-boots, boots, boots for sales and marketing.
60% of respondents Accenture executives also recognize that the resources of their organizations is not sufficient for the development and implementation of the assistant on their own, and 54% prefer for such projects to cooperate with external developers.
Forward into the future
To some extent, the future has come imperceptibly, but completely we realize this after the chat bots will become an accepted standard, such as mobile communications or the Internet. The future of chat-bots as a single ecosystem around the person. This virtual assistant is available anytime and anywhere, understanding normal conversational speech and which offers a single channel for all services. In practice, this would look like this: using system smart home the user within a single dialogue will be able to communicate with the Bank, to clarify financial reports, order food delivery or shopping, make a Bank transfer, etc.
However, the chat-bots are also beginning to consider the preferences of the user as the experience, to do the communication with the user more personalized, knowing his habits and features, also read the emotions of the user and predicting their desires and needs. Then it is time to talk about the emergence of a completely different solution, let me in which chat bots and virtual assistants were only intermediate stages.
Read more •••