Today, conversational interfaces are increasingly evolved and, thanks to Artificial Intelligence, they allow human-machine interaction to simplify many daily activities. With Pigro the company can query the knowledge base in natural language, choosing the solution that best suits its needs: interface, chatbot or API.
Conversational interfaces, also known as CUIs, belong to the broader field of user interfaces (UIs), which are systems designed to enable human-computer interaction.
The definition of conversational interface encompasses all those applications that allow the user to ask questions in natural language to a software, that is, using the language used by humans to communicate verbally or textually. In fact, within this set you can find chatbots, virtual assistants, bots, conversational agents, voice assistants, voicebots, etc.
CUIs replicate a conversation between humans on digital channels and have increasingly emerged as a tool to facilitate interaction between companies and customers, making it fluid thanks precisely to the ability to use natural language.
The interaction with the user can be both written and oral and the resulting dialogue can be more or less complex: conversational interfaces can, in fact, be based on the use of Artificial Intelligence, which allows the processing of natural language.
Thanks to AI it is possible to analyze the language and return a response consistent with the impulse received by the user, making them actually able to entertain a conversation.
In particular, the so-called Machine Learning (or machine learning) can be used in the design phase, to build algorithms capable of making the interface more precise in providing answers to the user.
NLP (Natural Language Processing) technology, on the other hand, allows computers to analyze and detect human language patterns, based on the syntax of the language: in a few words, they analyze the language and, by mimicking it, return a response as consistent as possible with that impulse.
In this way, it is possible to provide a user experience as close as possible to a human conversation, more perceptive of behaviors and intents and, consequently, more productive.
AI is used both at the information storage stage, to understand what the user is asking, and later to build and provide an answer to the question, due to its ability to analyze data and use it to make decisions in real time. In addition, many of these conversational systems are constantly continuing to learn and develop larger bases of content, thanks to the huge many databases and data sources available on the web today, which also allow for improved self-learning performance, based on the questions asked and feedback provided by the user.
Some examples of conversational interfaces can be chatbots, voicebots and virtual assistants, used both in first-level support, which requires easy actions and low-complex interactions to solve simple requests, and in purchase or booking support experiences, as in many e-commerce or travel sites, which require a greater ability to handle long and complex interactions.
The first conversational interfaces were born around 1960, with the so-called “text-based” dialogue systems, which aimed to answer questions, and chatbots that simulated casual conversations.
Since the 80’s, the idea of conversational interface as a system with which humans can interact for a variety of different purposes is gaining ground; for this reason, research in various fields is focused on the study and development of such systems.
One of the first conversational interfaces ever developed was the chatbot Eliza, created between 1964 and ’66 in the Artificial Intelligence lab at MIT (Massachusetts Institute of Technology).
Eliza was created for the purpose of filling the role of a psychotherapist. Its creator wanted to show how the lack of depth of character affected the relationships between people and machines, but many individuals who came into contact with Eliza attributed to the chatbot feelings similar to human ones.
Later, an example of a text-based chatbot, like Eliza, was Parry, created in 1972 and given the personality of a schizophrenic patient.
Further examples of interfaces designed to interact with users were Jabberwaky (1988), ALICE (1995), and the better known Clippy, the “paperclip” agent for Microsoft Office.
In 2006 IBM presents Watson, an artificial intelligence system developed from the idea of answering questions in a TV quiz. In 2013, the same company announced the first commercial application of Watson, which will be used in decision management in lung cancer treatment at MSKCC, a cancer research and study center, located in New York.
This was followed by projects of interfaces similar to IBM’s, including the now famous virtual assistants from Apple (Siri), Amazon (Alexa), Windows (Cortana) and Google Assistant. To date, it is not strange to have a smart device in the home, with which we interact using our voice and, in many cases, this technology is also used by our mobile devices. The purpose is to simplify everyday actions, such as setting a timer, checking the outside temperature or locating a place and knowing the route.
These artificial intelligences are also commonly used in many websites, by companies in a wide variety of industries to facilitate and manage relationships with users: thanks to the use of chatbots and virtual assistants on messaging apps and social, it is possible to manage customer support, from assistance to purchase, entertainment, to booking a vacation.
This type of interface is usually also called “flow chatbot“, since it operates according to predefined rules. For their development are, in fact, built predefined dialogue paths, which can guide the user to perform certain actions during the interaction, which typically occurs with buttons and keywords, rather than freely typing a command.
So-called chatbots, or text-based web or mobile systems, are interfaces that rely on an impulse sent by the user in the form of written text to provide a response, which can be written or spoken.
Given the large availability of data in written form (which they draw on), these types of interfaces are faster to implement. Depending on the type of technology used, chatbots can learn information categorized by keywords, tags or specific terms, as is the case with NLU – Natural language understanding. According to this approach, the information used to build the chatbot’s knowledge base must also be analyzed and “understood” from a semantic point of view. This results in a longer time for the implementation of such systems.
For interfaces that are based on a statistical AI approach, the implementation time will be faster since it is based on the correlations present in the text (e.g.: question and answer pairs that the chatbot extracts from the documentation base). Therefore, this type of interfaces does not require the user to use specific terms to perform the search. For further information: “Artificial intelligence with a statistical approach“.
The information retrieved can be of a generic nature, as in the case of the Google assistant, which opens a dialog box for searching the web, or more specific, such as a portion of text, or a specific service.
Similar to voice interfaces, they clearly differ in the type of visual front-end that the user uses to receive information.
These types of interfaces are useful in cases where the company needs to convey more complex information to the user, as they can have the aid of text, links, and graphics.
Voice-based conversational interfaces are systems that allow the user to complete an action by uttering a command.
Siri, launched in 2011 by Apple, was one of the first widely adopted voice assistants, initially available to all Iphone owners and later integrated into home devices as well. Numerous voice assistants, such as those belonging to Google and Amazon, were then developed with the aim of making users’ homes connected: thanks to the use of “smart” devices, they allow a whole series of actions to be carried out by pronouncing a simple voice command.
Following this, great progress has been made, especially because this type of interface has been used extensively in the e-commerce sales sector, to ensure fast and effortless user interaction. A limitation is, however, represented by the lack of text and graphics: while for some simple actions, such as re-ordering an already known product, voice is sufficient, for others, such as examining a new article or choosing an item from a menu, this type of interface is less suitable.
Hybrid interfaces are made up of mixed type interactions with the user, who will have both the ability to type freely and interact with the chatbot, but in some cases can be guided in performing certain actions with selection buttons and keywords.
The most common conversational interfaces today are natural language interfaces (such as chatbots and virtual assistants), which possess some key features.
Some of the most common features of natural language-based conversational interfaces are listed below:
– the ability to interpret natural language, understanding user requests;
– the ability to recognize the fundamental elements of a text and to subdivide them into categories, such as places, numbers, people, etc.; and
– ability to analyze syntax, hence the ability to relate the elements of a sentence and phrases in a period to each other to create a discourse;
– ability to correct grammatical errors, such as occurs in prompts while typing;
– ability to learn from interaction and context, so as to be able to suggest coherent solutions to the user or anticipate subsequent requests.
Conversational interfaces can be used in these areas in a variety of ways, both to facilitate the work of staff (internal support and help desk) and to support the performance of services and information retrieval for the end customer (customer service and customer care functions).
The main application areas of CUI technologies are:
Un altro campo di applicazione è quello dell’assistenza al cliente, come nel caso di help desk che permettono di risolvere problematiche specifiche o di veri e propri agenti virtuali che suggeriscono opportunità di investimento personalizzate o altri servizi come polizze assicurative, accompagnando l’utente negli step per la sottoscrizione.
– Financial and insurance services
Banks, credit institutions and insurance companies use chatbots to facilitate certain operations on their websites and customer counters, or to provide information on services provided.
Some examples are information retrieval, as in the case of user profiles associated with a bank, where the chatbot provides information about the bank account, transactions and, in some cases, allows new ones to be made via instant messaging applications.
Another field of application is that of customer assistance, as in the case of help desks that allow to solve specific problems or real virtual agents that suggest personalized investment opportunities or other services such as insurance policies, accompanying the user in the steps to subscribe.
One of the areas of greatest use of conversational platforms is represented by e-commerce sites and online sales.
In these cases, the chatbot can have a product search function, allowing a more natural and fluid use of product catalogs, or a marketing function.
In this case, the interfaces are used to suggest to the user products or services to add to the cart, promotions and offers and provide a personalized experience, based on his preferences, products already purchased or pages viewed.
Finally, after making a purchase, the customer may often need to view the status of the shipment, download the invoice or even inquire about return policies: chatbots can provide assistance for all necessary post-sale operations.
– Human Resources
In the field of human resources, the use of chatbots is emerging more and more to handle requests inherent to the normal management of employees, such as the issuance of payroll, counting of leave and vacation, etc..
In addition, it is increasingly common to use these interfaces in the recruiting phases of candidates, to perform an initial screening of resumes, presetting some basic fundamental characteristics that the artificial intelligence of the chatbot will track down among the candidate’s qualifications.
– Public Administration
Still not very widespread, but certainly useful, conversational platforms can support citizens in reaching public administrations, as in the case of chatbots on institutional sites, to provide information and deliver services. On the other hand, on the side of employees, they can streamline repetitive activities by assisting them, for example, in carrying out paperwork and filling out documents.
Artificial intelligence solutions such as conversational interfaces are increasingly being used in healthcare to support communication with patients, in particular diagnostics and monitoring, as well as information retrieval. Especially in health emergency situations, it is essential to be able to deliver services and facilitate doctor-patient communications even at a distance.
– Entertainment, tourism and information
A further use of chatbots is in the entertainment and media sector: starting from public information services, such as newspapers, which provide suggestions on the most suitable content to be consulted by the user. Also in the world of entertainment, chatbots are used to provide information or facilitate the purchase of tickets for events, concerts, cinemas, theaters, etc..
In the tourism sector, then, the use of conversational assistants constitutes a fundamental element for customer assistance, who can not only receive information on itineraries, but also book trips, stays and travel without physically going to a point of sale.
– Logistics sector
Solutions such as chatbots can be valuable allies when it comes to providing information to users on everything related to operations, warehouses, goods movements, shipping times, order changes, etc.. And it is precisely the ability to respond quickly and be available 24/7 that is a differentiator in this field.
– Operational sector
Voice assistants can be used on construction sites and in the field by operators who need to know how a machine works, or to record certain operations as they are carried out. The ability to use voice to interact with the interface is a significant time saver, as it allows the assistant to be consulted even when engaged in manual tasks.
– Operating systems – integrated applications
Well-known is the case of virtual assistants integrated into operating systems such as iOS’s Siri, or Windows’ Cortana. These voice assistants are integrated into mobile and desktop devices that support such operating systems and allow users to access applications and device features (calls, messages, weather, news, etc.), through the use of voice commands.
The use of conversational interfaces has become more and more frequent in the business world, also given the amount of advantages that can derive from the use of these tools that have become, to all effects, allies in working activities.
Some of the main points in favor of AI-based CUIs can be summarized as follows:
– as opposed to humans, they are available 24/7;
– they respond very quickly and simultaneously to a large number of users;
– they are able to mimic human behavior (in fact they are artificial intelligence), improving the user experience;
– can be integrated into any type of service and are multiplatform, to adapt to the needs of use;
– they are able to automate repetitive work: all those activities performed frequently can be performed by the chatbot, reducing the possibility of errors;
– they can use multiple communication channels, such as text, audio, images;
Consequently, many problems inherent to productivity and customer satisfaction can be solved, thanks to the introduction of conversational interfaces in the business environment:
– Enhance work and increase productivity: the use of interfaces such as chatbots can support employees in their search for information and decrease repetitive tasks, as in the case of Help Desk agents in technical support requests. If you have all the information you need to do your job quickly, you’ll be able to reduce the workload for all business units, which can then focus on higher value-added activities;
– Transforming the Customer Experience and Increasing Sales: especially in the B2C sector, the use of chatbots and conversational solutions helps to improve the user experience, to whom personalized suggestions are made that guide him in the purchase path. Moreover, even with regard to assistance and Customer Care, a chatbot can be a quick and effective solution to answer simple customer requests, improving the customer’s perception of the company;
– Reduce costs: by automating low-level tasks and repetitive actions, such as administrative activities, updating data in CRM, generating documents and consulting them within the company’s knowledge base, conversational interfaces also save costs related to the use of personnel to perform all these activities;
– Facilitating access to corporate knowledge and Knowledge Sharing: conversational platforms based on artificial intelligence can also be used in the area of Knowledge Management, which enables the management and organization of all corporate intellectual capital. Conversational interfaces can be used as true search engines in documentation, providing a centralized access point to knowledge and helping to disseminate information among employees;
– Facilitate onboarding and circulate company culture: conversational solutions such as chatbots can also be used for outreach purposes, for example to answer new employees’ questions about particular company policies and procedures. This will make it easier for the HR department to onboard and train new employees, helping to circulate information and spread company culture values.
Some of the key features that emerged from the opinions of CUI experts and leaders relate to both the design phase of interfaces and the focus on how the user will interact with them.
To talk about challenges, it is useful to introduce the concept of “pervasiveness” and “information architecture.”
The latter indicates precisely the design of information systems, software, websites, intranets, online communities based on usability and availability, therefore as usable as possible for users.
At this point we can say that an information architecture is pervasive when it becomes usable by the user on multiple channels and through different modalities.
The “multi-channel“, also mentioned by Henry Jenkins in his book “Convergence culture”, is the characteristic that concerns a service or a system, which can be used by the user through different channels (for example, web site, application, telephone call).
At this point, it is useful to introduce the fundamental difference with the concept of “cross-channel“, which instead concerns the use of multiple channels to complete a fruition experience. The user will complete part of his journey on one channel, part on another, since they offer different experiences and functions, but are complementary for the achievement of the user’s objective.
Regarding, in particular, conversational interfaces, a striking example of these concepts is provided by Peter Morville, father of information architecture.
Referring to a well-known and commonly used voice assistant, he notes how the impossibility of using multiple channels (auditory and visual) represents an obstacle for the user, whose experience is relegated to the use of voice alone, for example when searching for information.
Hence, one of the design challenges is to create interfaces that mix audiovisual input and output to enable better visualization and search, thus creating a cross-channel information architecture.
To understand what challenges the development of conversational interfaces will still face, it may be useful to take the point of view of the consumer, the one who interfaces (daily or almost daily) with chatbots, virtual assistants and conversational agents.
If, on the one hand, the user is more likely to interact with a robot, since it is always available and ready to respond, on the other hand, the level of empathy in the conversation will never reach that of a human dialogue.
For this reason it will be necessary to:
– find the right balance between human and robotic interactions to stimulate greater engagement;
– equipping conversational assistants with additional features, such as images and videos
in order to improve the user and customer experience, creating more engagement and offering better services than a common customer care.
In addition, in the design phase it is useful to give the interface a “personality”, for example by establishing the tone of voice that it should adopt, just as you do with the character in a story.
For this reason, it is fundamental to connect the study and design of the interface with market analysis that identifies the reference target: knowing how it acts and what it thinks makes it possible to understand what type of interaction it will create, what suggestions to introduce in the dialogue, so as to anticipate the intentions of the user and also encourage self-learning on the part of the AI.
As for the business side, certainly one of the critical points to be addressed in the development of artificial intelligence technologies (including CUI) is to understand how to introduce these innovations and which areas to involve.
Certainly, it will be necessary to start with the design of the user experience, then develop architecture and technology, all the way through to legal and compliance issues.
Investing in these areas to introduce a conversational interface will be easier if use cases for conversational assistants are identified (see Chapter 5.Uses and Applications).
Addressing these challenges in the not-too-distant future will lead to being able to solve critical issues noted by both consumers and businesses and use natural language interfaces as a tool to increase audience engagement and build customer loyalty.
In the framework of digital transformation of public administration, carried out in the UK by the Government Digital Service (GDS), founded in 2011, the potential offered by artificial intelligence represents valid tools to be exploited to digitalize services, facilitate communication, and the work of public offices.
Implementing a Knowledge Management system represents a support for all the activities that make up the IT Services management strategy, which we talked about in the last article. In fact, KM ensures that the right information is always available to do the job and, more importantly, to facilitate the decision-making process.
The objective of ITSM is to relate the Information Technology activities carried out by an organization to all stakeholders, i.e. all those figures between business, customers, and users.