How can AI and machine learning improve user experience?


What can AI and ML do to improve customer experience? Since the beginning of online shopping, AI and ML have been closely involved in the entire online shopping process. If you don’t get shopping advice, you may not be able to enjoy the high-quality services of Taobao or any other shopping website. These suggestions are usually personalized matching based on the supplier’s understanding of your features, including your purchase history, browsing history and more. Taobao and other online businesses want to create a digital version of a salesperson who knows you and your tastes, and can guide you to use the products you buy without error.

Everything starts with good data.

To achieve this vision, we need to start with some heavy work at the back end. Who is your customer? Do you really know who they are? All customers leave a data path, but the data path is a series of fragments, and it’s really hard to correlate these fragments. If a customer has multiple accounts, can you find out? If customers have separate accounts for business and other purposes, can you associate them? If an organization uses many different names, can you find that they are actually a single organization? Customer experience begins with an accurate understanding of who customers are and how they relate to each other, erasing customer lists to eliminate duplication called entity parsing, which used to be an area for companies with large amounts of data. Over time, we can see that entity parsing solutions are democratizing: a large number of start-ups have emerged that provide entity parsing software and services for small and medium-sized organizations.

Once you know who your customers are, you have to ask how well you know them. A comprehensive understanding of customer activities is very important for understanding the needs of customers, such as what data they have and how to use it. ML and AI are now widely used as tools for data collection: processing data streams from sensors, applications and other sources. Collecting customer data can be intrusive and morally problematic, so when you build an understanding of your customers, make sure they agree and do not compromise their privacy.

ML is not fundamentally different from any other type of computation: the rule “Garbage in, Garbage out” still applies. If your training data is of poor quality, your results will be poor. As the number of data sources increases, the number of potential data fields and variables will also increase, and errors may occur, such as transcription errors, printing errors and so on. In the past, we could correct and repair data manually, but manual correction of data is an error-prone and tedious task, and takes up the time of most data scientists. Like entity parsing, data quality and data repair have become a hot topic in recent research, and a new set of machine learning tools for automatic data cleaning has emerged.


A common application of machine learning and AI to customer experience is personalized recommendation system. In recent years, hybrid recommendation system, which combines multiple recommendation strategies, has become more and more popular. Many hybrid recommender systems rely on data from many different sources, and deep learning models are usually part of such a system. Although most of the existing models are deployed after training, advanced recommendation and personalized systems are real-time. Many companies begin to use reinforcement learning, online learning and personalized algorithms to build recommendation systems to continuously train models against real-time data.

Machine learning and artificial intelligence can automatically perform many different enterprise tasks and workflows, including customer interaction. At present, there are “experienced” chat robots on the market, which can automate all aspects of customer service. So far, chat robots have not reached the level of human beings, but if well designed, simple “common problems” robots can bring good customer conversion rate. We are at the early stage of understanding, but in the past year, we have seen many breakthroughs. As our ability to build complex language models improves, we can see the progress of chat robots in many stages: from providing notifications to managing simple question-and-answer scenarios, to understanding context and participating in simple dialogues, and finally to “understanding” the needs of users by personal assistants. With the improvement of chat robots, we hope that they can become an indispensable part of customer service. In order for chat robots to achieve this level of performance, they need to integrate real-time recommendation and personalization, and they need to understand customers and human nature.

Fraud detection is another technology that is applying machine learning. Fraud detection involves ongoing battles between the good and the bad, and fraud experts are inventing more sophisticated online crime technologies. Fraud is no longer human-to-human: it’s automated, like robots buying tickets for concerts because they can be sold again. As we have seen in many recent elections, criminals can easily infiltrate social media by creating a robot full of automatic responses. It’s really difficult to find these robots and stop them in real time. Only machine learning is possible. Even so, it’s a difficult problem to solve.

Advances in voice technology and emotional detection will further reduce friction in automated customer interaction. Multimodal models that combine different types of input (audio, text, visual) will make it easier to respond appropriately to customers; customers may be able to show you what they want or send real-time videos of the problems they face. Although the interaction between humans and robots often puts users in the creepy “Mysterious Valley”, it is certain that future customers will be more familiar with robots than we are today.

But if we want our customers to pass through the other side of this mysterious valley, we must also respect their value. AI and ML applications that affect customers must respect privacy, they must be secure, they must be fair and impartial. These challenges are not simple, but if the customer eventually feels abused, the technology will not improve the customer experience.

What will machine learning and AI do for customer experience? It has done a lot. But it has more to do, and what it has to do is to build a more natural customer experience in the future.

Author of this article: The tiger says eight things

Read the original text

This article is the original content of Yunqi Community, which can not be reproduced without permission.

Recommended Today

Nltk natural language processing library

Natural language processing, usually referred to as NLP, is a branch of artificial intelligence, dealing with the interaction between computers and people using natural language. The ultimate goal of NLP is to read, interpret, understand and understand human language in a valuable way. Most NLP technologies rely on machine learning to extract meaning from human […]