Chatbots use machine learning

Artificial Intelligence

The first dialog systems were developed and used as early as the mid-1960s and early 1970s. They were based on rigid rules and followed predetermined processes. Even today, such simple "bots" play a role when it comes to processing largely standardized processes.

Typical use cases include notifying changes of address, entering a SEPA mandate or reporting damage to an insurance company. The questions and answers of the bot system are fixed in such rule-based dialog processes and are usually created manually in the corresponding software. In an extended form, these simple bots can be connected to subsystems in order to exchange the recorded information.

Such dialogue systems may be useful in simple, manageable tasks, but they have some serious disadvantages: If, for example, the input of the human dialogue partner does not correspond exactly to a stored pattern, the bot cannot do anything with it. Branches and variants in the course of the dialog must be precisely foreseen and defined. With every change in the process chain, the dialog branches of the bot must also be adjusted.

Rule-based systems are therefore particularly suitable for processes in which questions and answers are in a direct one-to-one relationship. They fail in more complex scenarios in which the bot needs additional information from the dialogue partner, for example, in order to be able to answer a question.

How a bot becomes intelligent

Modern dialog systems are designed to be much more flexible. You can communicate with people in natural language and do not need a rigid flowchart. In order to generate such language assistants, the following machine learning functions are used, which are mostly implemented with the help of neural networks:

- Recognize and output language: Voice assistants first convert spoken language into text (Speech-to-Text), analyze it and give the answer in natural-sounding language (Text-to-Speech).

- Detection of intentions: In order to be able to give an adequate answer, the bot must recognize the content of the (spoken) text and deduce the intention of the dialogue partner from it. This analysis takes place via NLU modules (Natural Language Understanding), which are also based on neural, self-learning networks.

- Find and process information: Based on the recognized intentions and the content of a dialogue, ML-based systems based on neural networks can independently search, process and make available information in knowledge databases or logistics systems. Answers are not fixed, as is the case with the rule-based variants. Instead, the bot reacts flexibly to the question and, if necessary, compiles the answers from various sources. He can also actively inquire if important information is missing to answer the question. If there are several possible solutions, the bot can evaluate them and classify them based on their probability.

- Proactive promotion extension: Based on the dialog history, current bot systems can recognize the areas in which response options or branches are missing, suggest new intents to be modeled and identify missing information.

How to make your own bot

Before creating a bot, there should be a requirements analysis. It defines which specific tasks the digital assistant should take on. To do this, you should collect as much information as possible. Existing conversation guides from call centers and support departments or interviews with employees in the departments concerned are helpful, for example.

It is also advisable to work with service providers who have in-depth experience with the industry or department-specific processes to be modeled and who can offer pre-trained bots for the respective issue.

In addition, it is advisable to coordinate with IT and operations at an early stage and to clarify which systems the bot must be integrated into and from which sources it obtains information. Legal and regulatory issues also need to be addressed. Such specifications can, for example, be decisive for the question of whether a bot can be operated in the cloud or not.

On the basis of the process chains to be implemented and the coordination with the departments, so-called entities are now defined - units of knowledge that the bot needs to answer a question and which it may have to actively query. The actual bot development depends on the technology platform chosen. In graphically based solutions, dialog systems can be put together very quickly, even by employees without programming knowledge. They are recommended if users from the specialist departments should continue to look after the bot. Others are based on pure code and can therefore only be operated with appropriate experience.

Once the bot has been completed in an initial version, the training phase begins. In principle, a distinction can be made between two forms of learning: The bot can independently differentiate between good and less good answers based on user reactions and thus increase its hit rate. For this purpose, the feedback should be in an easily interpretable, machine-readable form, for example as a rating star or scale. Free text comments require a human curator to translate user responses into actions for bot training.

Alternatively or in addition, the bot can be adapted and improved directly. On the basis of previous conversations, experts identify problem areas and process chains in which the bot was unsure, for example, or in which the dialog was broken off or escalated to a human contact. By analyzing the conversations, targeted changes can be made to the bot control, such as requesting additional information or changing dialog processes.

If the bot is wrong

The goal of process automation is usually that the bot can process a case conclusively. While this is possible without any problems when ordering a pizza, legal questions arise in the insurance and banking environment, but also in other industries: For example, who is liable if a bot makes the wrong decision, for example because it has drawn incorrect conclusions from the training data?

It is therefore advisable in such cases to use bots to prepare decisions and to leave the completion of the transaction to a human employee. This has several advantages: On the one hand, one avoids the liability discussion, on the other hand, the bot's decisions can be checked directly and, if necessary, improved through further training. In the best case scenario, this approach builds up so much trust that the bot can act independently after a transition phase.

Conclusion

Rule-based bots have been known for many years. You work through rigid dialogue guidelines and have little to do with AI. Even learning systems based on neural networks make it possible to create voice assistants and chatbots that interact with people in a much more flexible and natural way. These "intelligent" bots recognize questions in natural language, extract a speaker's intentions, and combine information from various sources. In this way, the answers of the machine conversation partners not only get better and better, they can also independently discover new fields of action and thus expand their area of ‚Äč‚Äčapplication. (mb)