Chatbots, Voicebots, AI for Business in 2020 – 5 Key Trends

Over the past 12 months, the use of bots has been a hot topic in business. Not all predictions about the effects of intelligent automation have come true, but 2020 promises even more intensive experimentation with incorporating these solutions into customer service, e-marketing, and e-commerce strategies. As Gartner predicted, artificial intelligence (AI) in the coming years will deepen the customer experience – 47% of organisations will use chatbots for customer service, and 40% will implement virtual assistants. It's already happening.

According to Global Market Insights, the total global chatbot market will exceed $1.3 billion by 2024. Artificial intelligence is reorganising the way companies communicate both internally and with customers – AI is crucial for automated learning processes and flexible interpretation of automated business communication.

More and more industries intend to use virtual assistants to automate business processes in order to ensure the highest quality of customer service and improve consumer experiences. Company communication with customers is changing radically.

Thinking About Virtual Assistants Supporting Business – What's Worth Paying Attention to in 2020?

1. Higher Conversationality, Smoother Dialogue – Chatbots Learning from Tons of Data

Until now, bots went for quantity – deployed in numbers of several hundred thousand, in customer service or on fan pages – they delivered average conversation quality. The most common reason was the use of the simplest, free platforms for building them, lacking more advanced AI and ML (machine learning) mechanisms. After the false start, many brands invested in both technology and building competent teams, permanently developing UX (user experience) and training the bot. The first techniques to be applied are topic analysis – i.e., the scope of the basic knowledge base that the bot starts with – one of these techniques is topic modelling.

Recently, the number of available data resources and libraries has been growing, thanks to which the understanding (supported by NLP – natural language understanding) of virtual assistants is significantly improving. Thanks to developing algorithms, they are expected to understand customer questions and intentions much better.

The scope of application is expanding – modern chatbots not only support customers in purchasing processes, they can also, for example, take part in debates on social issues – as in IBM's Debater project.
What's more – Google's latest chatbot – Meena – was created based on 341 gigabytes of data from social media. Meena was programmed to imitate the way people communicate with each other – casually chatting about films, weekend plans, travel, hobbies – literally about anything that comes to mind. It makes a really impressive impression.

However, certain challenges remain challenges – among the main ones are:

  • Misunderstanding customer queries – chatbots often misinterpret user requests because they are unable to understand the customer's true intentions.
  • Misunderstanding dialogue details – chatbots often cannot interpret details in a dialogue, leading to misunderstandings.
  • Executing inaccurate commands – chatbots do not respond to technical commands issued by customers.

Difficulties in understanding dialects:

  • Chatbots are still unable to understand cultural dialects, which makes it difficult to understand intentions.
  • The user should no longer have to learn the chatbot; it's the chatbot that should learn what the user wants – this will translate into faster business processes and higher efficiency.

2. More Robots on Helplines

More Robots on Helplines

Due to technological progress in speech recognition, customers calling a helpline will no longer be sure whether they are talking to a consultant or a robot. Voice chatbots – voicebots – are a powerful tool for any contact centre, especially where the phone is the primary means of user contact (services for older people or those who cannot type while working).

Voicebots use automatic speech recognition, an innovative solution that enables converting speech to text and recognising intentions, keywords, and other elements of a query. This technology is a key element for voicebots and optimises helpline operations. Furthermore, a voicebot operates based on machine learning and AI algorithms, including Conversational AI technologies.

When a customer calls a helpline, the call is answered by a voice assistant – an algorithm that converts the customer's voice to text. The system then analyses the text query (via NLP – natural language processing) and matches an answer from the knowledge base. If the answer is in the base, the system – the voicebot – converts it to speech and presents it to the customer. If the knowledge base does not contain information matching the query, the system immediately transfers the call to a consultant.

3. Integration of Bots with Customer Data and CRM, Marketing Automation, ERP Systems – Supporting Business Processes

Integration of Bots with CRM Systems

Since Facebook, Microsoft, and Google launched APIs and platforms for building chatbots for businesses, most implementations have been based on simple question-answer solutions.

2019 initiated the second generation of bots – embedded in key business processes – and 2020 will only strengthen this trend. What does this mean? Bots will be able to resolve a customer's entire case, not just answer a reported problem. Integration with customer or product data is therefore key, enabling responses based on specific information. This requires integration and acquisition of data from CRM, Call Centre/Helpline, ERP, HR, and Marketing Automation software.

An example of such integration is the chatbot Decathlon Assistant by InteliWISE, which supports customers in the purchasing process, thereby relieving the company's contact centre. Thanks to InteliWISE's integration of Decathlon's internal systems, the chatbot enables, for example, easy checking of a specific order's status – simply by providing its number. It also helps find products for future purchases – thanks to connecting the chatbot software with the customer's product search engine, the customer searches for desired products in the store's resources and displays them to the customer in a carousel format. And for customers preferring in-store shopping – the Decathlon Assistant will find the nearest store – after entering the town's postal code or simply by checking the customer's location and their phone.

4. Searching for the Most Cost-Effective Applications in the Organisation

AI-based applications exist in HR and IT departments, but it is in customer service that chatbots have the greatest impact and are changing the way service is delivered. However, it is worth emphasising that many organisations have innovation centres with competent, entrepreneurial managers running them. This year, the trend of attempting to find new, previously unrealised applications of bots in the organisation will intensify.
Most organisations willingly use AI technologies. Intelligent applications are enterprise applications with built-in or integrated artificial intelligence that support or replace human-based actions – through intelligent automation, data insights, and recommendations aimed at improving performance and faster decision-making.

However, key to decisions about investing in technology will be understanding how much it actually costs to handle a given process by a human versus by artificial intelligence. And combining the work of qualified employees with AI support will bring the best results for business.

Source: The Economist February 22nd 2020

5. Regulators Will Take an Interest in AI, Including Chatbots and Voicebots

One of the important elements related to the use of bots in enterprises is becoming legal regulation – covering all of AI.

The European Commission is working on regulations for the digital economy – on the one hand, this concept is intended to regulate and encourage countries to work on AI, quantum computers, and digital data exchange, while on the other hand guaranteeing full security. The new digital strategy also assumes greater control over internet platforms (e.g., Facebook).
It is worth adding that in Poland, experts are working on their own AI strategy until 2027 – it assumes that Poland will be among the top 25% of places in the world working on artificial intelligence, and by 2025 over 700 companies dealing with AI will be established in the country.

Companies should not neglect legal issues related to artificial intelligence. They must be aware of potential regulatory risk and the risk of reputational loss or compensation payments. AI governance is the process of creating policies aimed at combating AI-related bias and the potential negative effects of such activities. Requirements for transparency of data sources and algorithms should be established to reduce risk and increase trust. To develop principles for managing AI-related risk, data experts should focus on three areas: trust, transparency, and diversity.

Key for many companies will therefore be control over the algorithm's operation and the ability to influence how it works. Companies have already started to require not only a working bot but also content management tools.

Source: InteliWISE S.A.

What Technologies Do We Still Have to Wait For?

Virtual assistants are taking the market by storm and operate based on increasingly advanced technologies, but we can already see that some predictions about AI were somewhat exaggerated – and 2020 will not bring major changes in certain areas.

Firstly, we will still have to wait for mass-scale purchase personalisation – yes, we increasingly buy using voice assistants, but this year is not the time for a shopping boom. These purchases are directly related to the functionality of voice commerce – i.e., buying using voice. We tell, for example, the Google Assistant the product we want to buy, and through a card linked to the service – the purchase simply happens – we don't have to additionally accept anything; the process is based only on voice commands. The technology is developing and reaching Poland too, but we will have to wait a while for its full bloom. Similarly with virtual assistants' understanding of context – they are learning more diligently and faster, but recognising differences in dialects and linguistic nuances is not their strongest suit. We'll see what 2021 brings.

More About InteliWISE S.A.

InteliWISE S.A. – a provider of AI-based software and cloud services (artificial intelligence, AI algorithms) for Polish and global clients, including those from the Fortune 500 list, the ranking of the largest American enterprises. The company supports customer service and digital business transformation, integrating intelligent automation technologies and increasing e-commerce and e-marketing efficiency. It has over 200 completed implementations of solutions such as Chatbot, Voicebot, Live Chat, and Contact Centre. InteliWISE is qualified in the TOP 10 global "Virtual Agent" technology providers, according to the independent analyst firm Forrester Research, and holds a US-patented technology for intelligent natural language recognition. InteliWISE is listed on the New Connect stock market (ticker: ITL), with strategic support provided by Intel Capital, and Asseco Group is a shareholder. More: www.inteliwise.pl

Summary

The article discusses five key trends in the use of artificial intelligence in business for 2020. These include: the development of chatbots based on advanced machine learning techniques, the growing role of voicebots on helplines, the integration of bots with CRM and ERP systems, the search for new applications in organisations, and the growing importance of legal regulations for AI.

Particular attention is drawn to the example of the Decathlon Assistant by InteliWISE, which serves customers in the purchasing process. The author also points out the limitations of the technology, such as problems with understanding dialects, and forecasts that the full development of voice commerce still requires time.