Artificial Intelligence in Practice
Page 8
It has also trained software “robots” to become so good at the strategy game Starcraft 2 that it can beat the computer team's AI bots on their highest difficulty setting. The interesting point here is that the computer team bots aren't really artificially intelligent in the sense we are interested in for the purpose of this book. They aren't controlled by self-learning algorithms but are simply programmed to repeatedly win – often by cheating.
Tencent's AI bots were able to challenge the Starcraft “AI” and defeat it by mimicking the tactics of the best human players. The system studied visual data from historic games, ingesting images at a rate of 16,000 frames per second, for two days, after which it was capable of defeating the computer AI at its highest level.5
Robots And Autonomy
Tencent has a strong interest in consumer robotics. In 2018, it led a round of investment in start-up UBtech, which specialises in robots for the home.6 As well as building bipedal robots that can move around in a human-like fashion and even negotiate stairs, it offers a range of machines that can fill various care, entertainment, companionship and home security functions.
It's also developing self-driving cars, cashier-less shops and home assistant speaker devices, in line with its competitors at home, Alibaba and Baidu, and in the United States, Amazon and Google.
Medical Technology
Where its work stands out above that of others, however, is in its application of AI in the medical field.
Tencent has integrated its phenomenally popular WeChat messaging platform with booking systems at 38,000 healthcare facilities. This means customers can use the app to book appointments online, as well as pay for their treatment through WeChat's payment systems.7
It has also partnered with the start-up iCarbonX, which uses genomics and advanced scanning technology to build up the most detailed digital models yet of individual humans, gathering data that will be invaluable in the development of precision medicines.8
This gives it access to a huge dataset logging patient interactions with the medical institutions which it can use to train machine learning models to predict demand for treatment across the country.
It also has a system that uses machine learning and computer vision to monitor the progress of Parkinson's disease, using only video footage of the patient. By measuring the patient's movements on camera, it can keep doctors updated on how they are progressing and assist with setting dosage levels. In many cases, this can reduce the need for patients to make hospital visits for routine check-ups for the condition.9
Tencent Miying – Artificial Intelligence In Hospitals
Tencent Miying is an AI medical imaging and diagnosing platform that Tencent has deployed in 10 hospitals in China, with agreements in place to extend to 100 more.10
It is comprised of two main systems – one that uses computer vision to assist doctors with the study of medical imaging such as MRI and X-ray scans, and one that assists with diagnosing and prescribing treatment.11
Miying is a product developed internally in Tencent's AI labs, and employs deep learning to power image recognition algorithms trained on thousands of scans. By learning to spot correlations between anomalies, which could be indicative of disease, it becomes an increasingly valuable tool to ease the workload of doctors in a country where they are in serious shortage.12
After completing analysis of its scans, the second AI function applies deep learning trained on thousands of medical documents and case notes to assist with diagnosing the patient and prescribing treatment.
The AI is trained to simulate the approach a doctor would take to study the data, although it is capable of doing it far more quickly and consistently. It can identify symptoms of more than 700 diseases.13
Key Challenges, Learning Points And Takeaways
Tencent is one of China's largest investors in AI and a business that continuously looks for opportunities to capitalise on AI across all of the industries in which it operates.
Its natural language processing, image recognition and machine learning technologies are all considered world leading, and expanding their use across its partner base and subsidiaries is a key business objective.
Its technology has huge implications for gaming in particular, where it can be used to verify identities and also create new gameplay challenges for players.
Tencent has also been recognised in particular for its success with building AI into healthcare systems, helping surgeries and hospitals to run smoothly, and assisting doctors with diagnosing and treating illness.
Notes
1Tencent, Tencent AI Lab: https://ai.tencent.com/ailab/index.html
2CB Insights, Rise Of China's Big Tech In AI: What Baidu, Alibaba, And Tencent Are Working On: https://www.cbinsights.com/research/china-baidu-alibaba-tencent-artificial-intelligence-dominance/
3CB Insights, Rise Of China's Big Tech In AI: What Baidu, Alibaba, And Tencent Are Working On: https://www.cbinsights.com/research/china-baidu-alibaba-tencent-artificial-intelligence-dominance/
4SCMP, Tencent employs facial recognition to detect minors in top-grossing mobile game Honour of Kings: https://www.scmp.com/tech/ big-tech/article/2166447/tencent-employs-facial-recognition-detect- minors-top-grossing-mobile
5The Next Web, https://thenextweb.com/artificial-intelligence/2018/09/ 20/tencent-created-ai-agents-that-can-beat-starcraft-2s-cheater-ai/
6Financial Times, Tencent drives China artificial intelligence push: https://www.ft.com/content/3143d482-4fc8-11e8-a7a9-37318e776bab
7CB Insights, Rise Of China's Big Tech In AI: What Baidu, Alibaba, And Tencent Are Working On: https://www.cbinsights.com/research/china-baidu-alibaba-tencent-artificial-intelligence-dominance/
8CB Insights, Lifting The Curtain On iCarbonX: China's Overnight Unicorn Is Attacking Everything From Genomics To Smart Toilets: https://www.cbinsights.com/research/icarbonx-teardown-genomics-ai-expert-research/
9The Week, How Tencent's AI can diagnose Parkinson's disease “within minutes”: http://www.theweek.co.uk/artificial-intelligence/96962/how-tencent-s-ai-can-diagnose-parkinson-s-disease-within-minutes
10Technode, How Tencent's medical ecosystem is shaping the future of China's healthcare: https://technode.com/2018/02/11/tencent-medical-ecosystem/
11Xinhua Finance Agency, Tencent releases first AI-aided medical platform: http://en.xfafinance.com/html/Industries/Health_Care/2018/ 361408.shtml
12Economist, China Needs Many More Primary Care Doctors: https:// www.economist.com/china/2017/05/11/china-needs-many-more-primary-care-doctors
13Xinhua Finance Agency, Tencent releases first AI-aided medical platform: http://en.xfafinance.com/html/Industries/Health_Care/2018/ 361408.shtml
Part 2
Retail, Consumer Goods and Food and Beverage Companies
11
Burberry: Using Artificial Intelligence To Sell Luxury
British fashion retailer Burberry sells its luxury goods online, but also through its network of over 500 bricks-and-mortar stores and concessions spread across 50 countries.1
In high-end fashion retail, customers still enjoy the personal touches that come with shopping in an exclusive boutique outlet. It also locates many of its outlets at airports where it caters to travelling clientele, who may need items immediately rather than delivered to their home in the near future.
Burberry's strategy for ensuring its actual stores remain competitive with virtual ones has been to bring many of the innovations made possible in the online world with web technology into the physical world – and much of this has been accomplished through artificial intelligence (AI).
What Problem Is Artificial Intelligence Helping To Solve?
Brands pushing the “luxury shopping” experience haven't been so quick to jump on board the e-commerce bandwagon as businesses aiming to replicate the day-to-day shopping experience.
Those buying expensive luxury goods clearly still appreciate being able to examine fine workmanship and high-end materials themselves before committing to buy, an
d enjoy being pampered in exclusive surroundings while they're doing it. This means a physical bricks -and-mortar presence in retail centers is a strategic necessity for some retailers, which is unlikely to be completely replaced by online.
Bricks-and-mortar stores still have some clear advantages over online-only outlets. Having an experienced customer service assistant on-hand is something that hasn't been fully replicated with AI yet. Although they are certainly on their way.
On the other hand, online shopping offers the convenience of not having to make a journey to the store, a world of choices and easy access to intelligent machine tools. The purpose of these tools (such as search engines) is generally to edit your potential selection down to something your human brain is capable of processing and making a decision on.
Retailers that depend on a physical storefront have the challenge of having to compete with this huge shift in choice and convenience that online retailers have brought to the game, or continue to lose customers to them.
How Is Artificial Intelligence Used In Practice?
To encourage well-heeled shoppers to keep coming into their stores, Burberry's mission has been to use advanced data technology – including AI – to replicate many of the advantages and conveniences we have come to expect online. This is chiefly carried out through a number of different loyalty programs.
Data that customers hand over is used to generate profiles and segment customers, so sales assistants can approach them with recommendations based not just on their own purchase history, but on thousands of others who fit a similar profile.
Data and AI are also used to understand why particular items may be selling well in stores but not so well online. One insight Burberry gained was the overriding importance of product images. When new images were uploaded for items that weren't selling well online, in one case sales increased by 100%.2
What Technology, Tools And Data Were Used?
With their customers’ permission, buying habits across the group's online and physical stores are tracked, and the generated data is made available to shop assistants via a tablet terminal. This means assistants can make suggestions and show customers products they might be interested in, based on their browsing history, past purchases and even analysis of their social media.
Burberry also uses radio-frequency identification tags on in-store products to give customers insights into how they were created, and ideas about how they can be worn. This also gives Burberry more information about what the customer is interested in, in the same way that Amazon will track which products its customers browse online.
What Were The Results?
Burberry is able to quickly build a picture of who the visitors to its physical stores are, and approach them with the same convenient recommendations they would get online. Those who are accustomed to old-fashioned customer service probably appreciate interacting with a sales assistant who knows their name too!
David Harris, Burberry's senior vice president of IT, said: “We believe that AI can deliver business value through making better products, faster, cheaper processes and/or more insightful analysis.”3
Key Challenges, Learning Points And Takeaways
Understanding, tracking and modelling customer behavior has been a feature of offline shopping for far longer than it has with online, thanks to loyalty programs. What's new is the ability to bring advanced AI solutions to the data – technologies developed to power the e-commerce revolution.
High-end fashion retailers need to keep a physical storefront for their discerning customers – AI can help replicate the convenience of shopping online in the real world.
Notes
1Statistica.com, Number of Burberry stores worldwide in 2018, by outlet type: https://www.statista.com/statistics/439282/burberry-number-of-stores-worldwide-by-outlet-type/
2Forbes, The Amazing Ways Burberry Is Using Artificial Intelligence and Big Data: https://www.forbes.com/sites/bernardmarr/2017/09/25/the-amazing-ways-burberry-is-using-artificial-intelligence-and-big-data-to-drive-success/#35325a4d4f63
3AI Business, Where are Burberry with AI? Exclusive Interview with David Harris, SVP of IT: https://aibusiness.com/where-are-burberry-with-ai-exclusive-interview-with-david-harris-svp-of-it/
12
Coca-Cola: Using Artificial Intelligence To Stay At The Top Of The Soft Drinks Market
As the world's largest beverage company, Coca-Cola serves more than 1.9 billion drinks every day, across over 500 brands, including Diet Coke, Coke Zero, Fanta, Sprite, Dasani, Powerade, Schweppes and Minute Maid.
Big data and artificial intelligence (AI) power everything that the business does – the global director of digital innovation, Greg Chambers, said: “Artificial intelligence is the foundation for everything we do. We create intelligent experiences. Artificial intelligence is the kernel that powers that experience.”1
What Problem Is Artificial Intelligence Helping To Solve?
Marketing soft drinks around the world is not a “one-size-fits-all affair”. Coca-Cola products are marketed and sold in over 200 countries. In each of these markets there are local differences concerning flavors, sugar and calorie contents, marketing preferences and competitors faced by the brand.
This means that to stay on top of the game in every territory, it must collect and analyze huge amounts of data from disparate sources to determine which of its 500 brands are likely to be well received. The taste of their most well-known brands will even differ from country to country, and understanding these local preferences is a hugely complex task.
How Is Artificial Intelligence Used In Practice?
Coca-Cola serves a large number of its drinks every day through vending machines. On newer machines, typically the customer will interact through a touch-screen display, enabling them to select the product they want and even customize it with “shots” of different flavours. The company has begun fitting these machines with AI algorithms allowing them to promote drinks and flavors that are most likely to be well received in the specific locations where they are installed.2
The vending machines can even alter their “mood” depending on where they are located – with machines in a shopping mall displaying a colourful, fun persona, those in a gym more focused on achieving performance, and those in a hospital appearing more functional.
Coca-Cola also uses AI to analyze social media and understand where, when and how its customers like to consume its products, as well as which products are popular in particular localities. With over 90% of consumers making purchasing decisions based on social media content,3 understanding how its billions of customers are discussing and interacting with the brand on platforms like Facebook, Twitter and Instagram is essential to its marketing strategy. To do this, Coca-Cola analyzed engagement with over 120,000 pieces of social content to understand the demographics and behavior of its customers and those discussing the products.
Another application of AI was in securing proof of purchase for the company's loyalty and reward schemes. When customers were asked to manually enter 14-digit product codes printed on bottle caps into websites and apps to verify their purchases, uptake was understandably low due to the unwieldy nature of the operation.
To encourage more customers to engage with these schemes, Coca-Cola worked to develop image recognition technology that allows purchases to be verified by taking a single smartphone picture.
What Technology, Tools And Data Were Used?
Coca-Cola collects data on local drink preferences through the interfaces on its touch-screen vending machines – over 1 million of them are installed in Japan alone.
To understand how its products are discussed and shared on social media, the company has set up 37 “social centers” to collect data and analyze it for insights using the Salesforce platform. The aim is to create more of the content that is shown to be effective at generating positive engagement. In the past, the process of creating this content was carried out by humans; however, the company has been actively looking at deve
loping automated systems that will create adverts and social content informed by social data.4
It also uses image recognition technology to target users who share pictures on social media inferring that they could be potential customers. In one example of this strategy in action, Coca-Cola targeted adverts for its Gold Peak brand of iced tea at those who posted images that suggested they enjoy iced tea, or in which the image recognition algorithms spotted logos of competing brands.5 Once the algorithms determined that specific individuals were likely to be fans of iced tea, and active social media users who shared images with their friends, the company knows that targeting these users with adverts is likely to be an efficient use of their advertising revenue.
For purchase verification, off-the-shelf image recognition technology proved to be insufficient for reading the low-resolution dot matrix printing used to stamp product codes onto packaging. So, Coca-Cola worked to develop its own image recognition solution using Google's TensorFlow technology.6 This used convolutional neural networks to enable machine recognition of codes that could often appear differently depending on when and where they were printed.
What Were The Results?
Analysis of the data from vending machines by AI algorithms allows Coca-Cola to more accurately understand how the buying habits of its billions of customers varies across the globe.