The Ma in Machine Learning

Machine Learning and AI in E-Commerce

Jack Ma’s Alibaba has been at the forefront of innovation since its founding in 1999 from an apartment in Hangzhou, China. Alibaba’s original but lasting mission has been to connect buyers from around the globe to Chinese suppliers. Today, Alibaba’s platform has over 617 million monthly mobile users and 552 active users on its China retail marketplaces. [1] This scale awards Alibaba unbelievable access to consumer data and the Company has sought to capitalize on this data by investing heavily in AI and machine learning as a core part of its current and future business strategy. Alibaba is the largest R&D spender in China and has invested in seven research labs focused on the business applications for AI and machine learning. [2] The investment is aimed to help the Company remain at the forefront of a continuously changing e-commerce landscape, both in China and globally. China is too supporting the mission, with plans to build a $1 trillion AI industry by 2030 [2].

Leveraging the constant influx of data from businesses and consumers on Alibaba’s platform, machine learning is able to identify regular patterns from the accumulated data to help make predictions and decisions to enhance the customer experience. [3] “These enhancements include personalized search results and shopping recommendations empowered by deep learning and data analytics, speech recognition and image analysis technology adopted in search functions, and intelligent customer service.” [4] By utilizing the power of machine learning to generate actionable insights for the business, Alibaba is able to showcase a constantly improving customer experience on the platform, creating more personalized e-commerce experience. As demonstrated, machine learning is more than a technological innovation; it will transform the way business is conducted as human decision making is increasingly replaced by algorithmic output. [5]

How Alibaba Uses Machine Learning

The use of machine learning is prevalent on the platform with Alibaba’s launch of its chatbot, Dian Xiaomi (store assistant). Built with AI deep-learning technologies and natural-language processing, Dian Xiaomi is the premier chatbot, answering product-related questions, offering personalized product recommendations, managing returns and refunds and offering discounts and promotions. [6] Dian Xiaomi has showed the ability to understand over ninety percent of customer questions, serving more than 3.5 million users a day. [6] The latest version of the chatbot can even sense a customer’s emotion and anxiety level and recommend a customer service agent intervene as necessary. If the need for a human response is triggered, the human representative helps direct the bot to the right answer or adds the answers to the database so that the chatbot can “learn” for next time. [6] The human element may ultimately become obsolete as the chatbot’s input data becomes increasingly complete.

Dian Xiaomi addresses a pain point for merchants on the platform. [6] With limited human resources, the chatbot was able to handle the heavy volume of customer inquiries during peak times, like during Singles Day, the largest shopping day of the year with over $31 billion in gross merchandise value. [7] Each merchant can adjust the workload that is allocated between the chatbot and customer-service staff, allowing each merchant to have more ownership of the customer service experience.

What’s Next?

How can machine learning scale beyond its e-commerce platform? Alibaba is leveraging its AI technology in the City Brain project, a project that started in Hangzhou that seeks to apply AI technology to traffic congestion. City Brain’s data was collected from a number of sources including direct monitoring of traffic, public transportation systems, mapping apps and cameras around the city. [8] Based on the data collected, Alibaba was subsequently given control over one hundred traffic light junctions and able to reduce traffic congestion by 15% in the city. [8] The collection and analysis of the data also allowed the city to respond faster to accidents as well as monitor illegal parking. [8] This application, in my mind, can utilize machine learning for increasingly better and automated results, to allow the machine to learn and dictate traffic patterns with a better feedback cycle.

The traffic issue is certainly not isolated to Hangzhou, as the findings and implications of this project are undoubtedly scalable to other cities around the globe that deal with significant traffic issues. Alongside the positive advancements in the project as well as with e-commerce come concerns over privacy and surveillance issues. The societal question becomes finding the balance between efficiency and privacy. What is the appropriate amount of data collection to promote an efficient, effective society but not infringe on an individual’s privacy? What is the appropriate body to regulate the use of this data? Can we apply the same regulations, once established, to the global society or do the rules need to be specific to regions / countries? Who decides its limits?

Word count (798)

[1] Laubscher, Hendrik. “For Brands, Alibaba Is The Gateway To China And Chinese Customers.” Forbes, June 2018. https://www.forbes.com/sites/hendriklaubscher/2018/06/28/brands-alibaba-is-the-gateway-to-china-chinese-customers/#51247d9e1658, accessed November 2018.

[2] Marr, Bernard. “The Amazing Ways Chinese Tech Giant Alibaba Uses Artificial Intelligence and Machine Learning.” Forbes, June 2018. https://www.forbes.com/sites/bernardmarr/2018/07/23/the-amazing-ways-chinese-tech-giant-alibaba-uses-artificial-intelligence-and-machine-learning/#559c4938117a, accessed November 2018.

[3] Alibaba Group. “What is Machine Learning Platform for AI?” https://www.alibabacloud.com/help/doc-detail/67461.htm, accessed November 2018.

[4] Alibaba Group Holdings Ltd. Form 20-F. Company Website, July 2018. https://www.alibabagroup.com/en/ir/secfilings, accessed November 2018.

[5] Zeng, Ming. “Alibaba and the Future of Business.” Harvard Business Review, September – October 2018. https://hbr.org/2018/09/alibaba-and-the-future-of-business, accessed November 2018.

[6] Wang, Susy. “Merchants Deploy Alibaba’s AI Customer-Service Chatbot.” Alizila, April 2018. https://www.alizila.com/online-merchants-deploy-alibabas-ai-customer-service-chatbot/, accessed November 2018.

[7] Hu, Krystal. “Chinese shoppers just spent a record $30.8 billion on Alibaba’s Singles Day.” Yahoo Finance, November 11, 2018. https://finance.yahoo.com/news/chinese-shoppers-just-spent-record-30-billion-alibabas-singles-day-175653137.html, accessed November 2018.

[8] Beall, Abigail. “In China, Alibaba’s data-hungry AI is controlling (and watching) cities.” Wired, May 2018. https://www.wired.co.uk/article/alibaba-city-brain-artificial-intelligence-china-kuala-lumpur, accessed November 2018.

 

Previous:

A Tech Titan in Healthcare: All Eyes on Deep Learning

Next:

Unlocking Efficiency Gains Through AI: Shell and Machine Learning

3 thoughts on “The Ma in Machine Learning

  1. Great post! I find it interesting that Alibaba is using its machine learning capability to improve traffic congestion. What does Alibaba get from this relationship? Are they compensated with money, improved delivery times, or awareness? I am also worried that with an integrated system for traffic management, or any other traditional public service, there may be the potential for external hacking, which could severely disrupt the flow of a city.

  2. Christie – You raised some interesting issues here, especially given the China context. With the impending implementation of the social credit system, Alibaba’s AI system could in theory be used by the Chinese government to monitor and assess the behavior of its citizens. For example, the chatbot Dian Xiaomi could be used to monitor enormous internet traffic (both the content and emotions displayed by internet users). A variant of the technology could be used to monitor the behavior and movement of citizens on the road. While this may not be seen as an issue in China (while the government aims to produce a harmonious society), there could be more resistance in other countries with priorities that are different than China.

    Also, it is inherently risky to provide so much personal data to a private company. We have seen in the case of Facebook and Cambridge Analytics the damages that can be caused by privatization of personal data, to both the public and the companies keeping hold of the data. Private companies that hold vast quantity of personal data should therefore have robust privacy protection policies and systems. Ideally, this should be governed and enforced by a public regulator. It is difficult to leave this job to another private company (which may have different incentives to the public) or individuals (who don’t have the necessary time, resources, expertise or public influence). In setting up the regulatory framework, the government and legislature should solicit public involvement as much as possible. This is a public interest issue after all; it should be up to the public where the line of balance falls. As such, each society may come out with a different set of rules and regulations, even if this means businesses will have to incur additional costs of implementing a different regulatory standard to what is otherwise common pool of data across the world.

  3. Very interesting read! I found it interesting that Alibaba created a chatbot to mitigate it’s limitations in human resources and handle periods of heavy customer inquiries. Do you think there could be any downsides to this? What happens if the data the chatbot is receiving becomes flawed? I’m worried about potential hacking, especially since Alibaba is one of the biggest and most powerful companies in the world. Overall, great post.

Leave a comment