Artificial Intelligence in Practice - Book review
1 January 2021
Dr. Moria Levy
"Artificial Intelligence in Practice: How 50 successful companies used AI and Machine Learning to Solve Problems" is a book by Bernard Marr in collaboration with Matt Ward. Published in 2019, the book sheds light on AI projects implemented by various companies, making it highly relevant and insightful for organizations familiar with the subject.
The book comprises a collection of 50 stories, providing summaries of each. It begins with a description of the 10 AI giants, followed by a detailed table outlining the main activities of all the other companies, concluding with a summary from the authors.
Organizations adopt AI to achieve three primary goals: 1) Transform how they understand and communicate with customers, 2) Provide improved solutions and services, and 3) Enhance and automate their business processes. Those organizations that succeed in adopting AI often witness a significant transformation in their approach to conducting business.
The book features leading AI organizations worldwide, including ALIBABA, GOOGLE, AMAZON, APPLE, BAIDU, FACEBOOK, IBM, JD.COM, MICROSOFT, and TENCENT. Each company demonstrates innovative and impactful uses of AI and Machine Learning.
Furthermore, the book includes a comprehensive table showcasing the activities of various companies across different sectors. The sectors covered include Trade, Consumerism, and Food; Media, Entertainment, and Communication; Service, Financial, and Health; Heavy Industry, Automotive, Aerospace, and Industry 4.0.
Having read this book, I was informed and amazed by the applications of AI and Machine Learning in diverse industries. I wholeheartedly recommend this book to anyone interested in the subject matter, as it serves as a practical guide for implementation and a valuable learning tool to explore the ever-evolving world of AI.
WORLD'S LEADING ORGANIZATIONS:
About the company: The Chinese company is the world's largest online purchasing giant.
• The company utilizes AI to facilitate online product sales, including:
- Personalized product prioritization based on individual buyer preferences.
- Customization of customer actions to enhance the likelihood of purchases during each visit.
- Implementation of chatbots for customer communication.
- Automated content creation for product descriptions.
• Alibaba also offers cloud services and AI tools to organizations interested in adopting AI, including:
- Machine learning environments for cloud customers.
- NLP (Natural Language Processing) tools.
• The company is actively involved in making cities "smart" by employing AI in various areas such as:
- Balancing traffic loads and optimizing street and city center lighting.
- Improving garbage collection processes.
- Implementing smart kiosks at train stations.
• In addition, Alibaba is engaged in smart agriculture, using AI to automate pig feeding and training decisions, taking advantage of China's position as the world's largest consumer and supplier in the industry.
ABOUT THE COMPANY: Google, or, in fact, Alphabet (Google's parent company), is known to us for its search engine, Waze, YouTube, cloud services, and more.
Across the company's various activities, including but not limited to:
- NLP request analysis
- Image recognition through deep learning
- Effective matching of results based on the understanding of the request through machine learning
• Self-driving cars
- WAYMO - the first autonomous car used for commercial purposes
- Automatic subtitle generation for movies
• Smart virtual assistants - Google Home
- Virtual assistants adapted to specific situations, such as booking an appointment at the hair salon
• Language translation - Improving translation through deep learning analysis of language components
• Medical Science Saves Lives - Learning medical scans of eyes for disease detection using deep learning.
About the company: Amazon is an online sales giant that initially gained fame for selling books online. It sees itself as a technology giant, competing with Google and Apple, and is constantly expanding into other solutions, such as Alexa, for home assistance.
AI activity: Amazon has integrated AI into all its company activities and operations for about a decade, using deep learning technology significantly. A unique concept for integrating data and technology between the company's various projects and departments enables learning from each other, facilitating continued expansion and product improvement:
• Personalized sales experience - Sales recommendations in multiple aspects.
• Warehouse Management - Use of robots equipped with deep learning for the selection and transportation of products.
• Personal virtual assistant - Alexa - Home Assist: Identifies when we wake up to start listening, understands, performs tasks, and learns from interactions.
• Cloud Services - AWS - Environment and tools that enable companies and organizations to implement AI.
• Drones - Drones used for transporting the products sold to people's homes.
About the company: Apple is the largest computer company in the world, engaged in many technologies ranging from MAC computers to iPhones, smartwatches, tablets, TVs, and more.
AI activity: Apple's AI activity is closely integrated into the devices it sells, making it a leader and pioneer in the field. The company excels in information security and can create a unique and engaging user experience:
- AI is integrated as a chip, starting from iPhone X, enabling deep learning for faster performance of various functions.
- Face recognition capabilities.
- Management of widgets (APPS).
- Empowering Apple app developers to integrate deep learning features.
- Infrastructure widgets that can identify objects the camera points to offer suggestions for improving images and more.
• Personal virtual assistant - Siri possesses NLP capabilities, including translation between 40 languages, ensuring precise execution of desired tasks, and more.
About the Company: Baidu is an internet services and products company based in China.
AI activity: Baidu incorporates AI into various aspects of its operations, including but not limited to:
- Improving search results.
- Image recognition capabilities.
- Video analysis.
• Self-driving cars
- Integrating sensors and algorithms while transmitting data to the cloud to identify road conditions and obstacles and provide recommended solutions.
- Apollo - an open environment for autonomous driving trucks on highways.
• Translations - Deep learning-based online translations between Chinese, Mandarin, English, and Japanese.
About the company: Facebook is a social media services giant offering extensive web services.
AI activity: The company leverages all the content it is exposed to as data to train and operate its machines. Their AI projects and products are designed to serve users in various ways:
• Customization for each user
- Personalized news feeds.
- Customized home pages and advertisements.
• Content control - Filtering out violent or false content, including fake news.
• Face recognition - Identifying people in photos.
• Understanding texts - DeepText - Comprehending the content users write and understanding their ideas and desires related to them.
• Suicide prevention - Listening to users and referring them to emergency services as needed.
About the Company: IBM is one of the largest computer companies, with a history of over 100 years, during which it held a leading position in the market.
AI activity: IBM's AI activity centers on Watson, a cognitive computing environment. Much of the activity involves customers purchasing the machine's services and developing projects either independently or with the assistance of IBM:
- Deep Blue: Chess (famously defeating the unknown champion).
- Victory in the Jeopardy game against world experts based on questions and answers in general knowledge.
• Chatbot at Royal Bank - Computerized customer service, with 40% of inquiries being answered by the machine.
• Medicine - American Cancer Association - Assisting physicians in recommending personalized drugs for cancer patients.
• Cosmetics - Estee Lauder - Development of new perfumes.
• Language processing - Project Debater - An infrastructure for language processing and dealing with complex issues.
About the Company: JD.com is a China Technology Trading Company.
AI Operations: JD.com integrates AI into all its shipping, logistics, and operations activities, providing significant benefits and efficiency:
• Computerized reservation center management - A reservation center with only four people manages about 200,000 orders daily, thanks to computerization.
• Robots for transportation and packaging - Smart robots efficiently transfer products to conveyors and handle packaging.
• Chatbot service user experience - The chatbot engages in conversations with customers and is claimed to offer its own words of poetry.
• Personalized advertisements - JD.com collaborates with huge online gaming companies in China for targeted and personalized promotions.
• AI-powered shipments - JD.com utilizes delivery drones and autonomous trucks for efficient and timely deliveries.
• Face recognition - Facial recognition technology is used for payments and package receipts, eliminating the need for wallets or IDs.
• Smart refrigerators - Refrigerators equipped with cameras and identification technology to manage inventory, missing items, expiration dates, and more.
• Smart Stores - JD.com's smart store's function without sellers, using facial recognition technology for payment and making purchases simply by looking at the camera and recognizing the face, without the need for a wallet.
About the company: Microsoft is a computing and technology giant that goes beyond personal and office computing, venturing into electronic consumerism, video games, cloud computing, and social media.
AI ACTIVITY: Microsoft's AI activity is guided by the democratization principle, aiming to provide extensive usability for users and enable companies to develop their AI based on Microsoft services and environments. Here are some critical aspects of their AI initiatives:
• Tools and development environments - AI Azure Cognitive Services
- Search capabilities.
- Speech recognition technology.
- Text analysis features.
- Image recognition, including face recognition.
- Translation services.
- AI School online platform for training in tool usage and robot building.
- Sketch2Code - A tool that converts drawings into HTML sites.
• Office 365
- Custom design features for users in PowerPoint.
- Word offers suggestions for meanings, equivalent terms, word corrections, and grammar and punctuation assistance.
• Cloud Services - Enhancing internet usage speed.
• Renault Formula 1 performance enhancement - In partnership with Renault, Microsoft works on improving performance in the Formula 1 racing domain.
About the company: Tencent is a Chinese giant offering internet services and technology, and it ranks among the most valuable companies in the world, primarily due to its social media activity and online gaming ventures.
AI ACTIVITY: The company shows aggressive investments in acquiring startups in the AI field, contributing to various sectors:
• Face recognition - Enabling commerce in entire cities through image-based (non-physical) digital cards.
- Identifying the age of users and applying restrictions accordingly.
- Company robots achieving victory in strategy games like StarCraft 2 against other bots.
• Home service robots
- Capable of moving around the house and performing tasks, including handling stairs.
- Engaging in various activities like entertainment, companionship, and security.
- Smart appointment scheduling and facilitating payments for medical services.
- Collaboration with iCarbonX and scanning technology for prescription drug development.
- Monitoring the progression of Parkinson's disease through patient observation, reducing the need for hospital visits.
- Assisting in analyzing MRI scans and providing appropriate treatment recommendations for over 700 diseases.
Sector: Trade, Consumerism, and Food
• Necessity - Selling luxury items and facilitating matches between sellers and buyers.
- Customer profile data.
- Purchasing habits data.
- Clothing display data.
- Details about the clothes offered.
- Sales data.
- RFIQ (Machine learning) technology.
• Need - Developing and selling drinks in self-machines and providing advertisements adapted to specific places and times.
- Consumption data related to Coca-Cola products.
- Social media data for marketing insights and consumer trends.
- Visual computing technology.
- NLP (Natural Language Processing) technology.
- Deep learning technology.
• Need - Focusing on improving the quality of pizzas produced, incorporating a virtual assistant for enhanced customer service, and utilizing autonomous vehicles for deliveries.
- NLP (Natural Language Processing) technology.
• Need - Understanding customers' needs and providing offers tailored to their situations, such as pregnancy plans.
- Network access data for understanding customer interactions.
- Club Customer Data for personalized marketing and loyalty programs.
- Operational Data for optimizing business processes.
- Machine learning technology for data analysis and customer insights.
- Improvement in service quality.
- Increasing sales.
- Adapting menus based on inventory, weather, and other factors.
- Predicting personal preferences to enhance customer experience.
- Data on purchases and orders made through kiosks for customer behavior analysis and preferences.
- Machine learning technology for data analysis and improving operations.
- Robots for efficient and automated processes in certain aspects of the business.
Need - Developing intelligent products for home, workplace, and industrial organizations, including implementing virtual assistants.
- Data from sensors in devices used to gather information and improve product performance.
- NLP (Natural Language Processing) for enhancing virtual assistant capabilities.
- Robotics technology for creating advanced intelligent products.
- Improving inventory management to ensure desired flavors in each branch.
- Increasing sales through personalized recommendations.
- Implementing deliveries using autonomous vehicles.
- Consumer behavior data for understanding preferences and trends.
- Data on total consumption based on location, date, weather, times, and more.
- Machine learning technology for data analysis and improving inventory management.
- Robots for efficient and automated processes in certain aspects of the business.
- Sending personalized fashion purchase offers to people based on predictions of their preferences.
- Understanding dimensions through image analysis to develop clothes that will be sold.
- Matching stylists to customers for personalized styling services.
- Individual consumer behavior data to understand preferences and trends.
- Total consumption data for analyzing overall customer choices.
- Machine learning technology for analyzing customer data and making personalized fashion recommendations.
Need - Recruitment process improvement through automated tests, video interview analysis, and expression analysis.
- Exam data for evaluating candidates' skills and qualifications.
- Video data from interviews for analysis and assessment.
- Company documents for reference and data analysis.
- NLP (Natural Language Processing) technology for analyzing written and spoken content during recruitment.
- Visual computing technology for video analysis and understanding facial expressions during interviews.
Need - Implementing ongoing live filling of shelves according to consumption patterns to ensure products are readily available for customers.
- Product photography data for identifying products and their characteristics.
- Shelving photos data for understanding shelf space and organization.
- Robotics technology, similar in application to autonomous vehicles, for automating the shelf stocking and replenishment process.
Sector: Media, entertainment, and communication
Need - Shortening queues for customers and enabling payments through wristbands worn by visitors.
- Cameras capture data, including crowd information and guest behavior.
- Appointment data to manage reservations and schedules efficiently.
- Parks perimeter data for security and crowd management.
- Machine learning technology for data analysis and enhancing guest experiences.
- Visual computing technology for analyzing camera data and understanding guest movements.
- Robotics technology for certain park attractions and experiences.
- RFID (Radio-Frequency Identification) technology for wristband payments and tracking.
Need - Reducing cyberbullying and creating a safer online environment for users.
- Content on the web, including user-generated posts and comments.
- NLP (Natural Language Processing) technology for analyzing and identifying potentially harmful or abusive language.
- Deep learning technology for content analysis and detecting harmful patterns or behavior.
Need - Linking employees to recommended jobs and courses and planning site functionalities for an improved user experience.
- Online content, including user profiles, job postings, and course offerings.
- NLP (Natural Language Processing) technology for analyzing and understanding user information and preferences.
- Machine learning technology for making job and course recommendations based on user data.
- Personalizing content recommendations for viewers based on their preferences.
- Improving data flow over the network for seamless streaming experiences.
- Analysis of existing viewing habits for many factors to predict user preferences.
- Deep learning technology, including audio analysis, for understanding viewer and content preferences.
- Video analysis using deep learning algorithms for content recommendations.
Need - Mechanizing the process of writing local news based on knowledge gathered by journalists.
- News worth covering, including current events and relevant topics.
- Resources on the web, including articles, reports, and other information sources.
- NLP (Natural Language Processing) technology for analyzing and processing news content.
- Machine learning algorithms for automating news writing based on gathered data.
Need - Providing personalized music recommendations to users based on their preferences.
- Analysis of existing usage habits of users to understand their music preferences.
- Song repository containing a vast collection of music.
- Machine learning technology for analyzing user data and making personalized music recommendations.
- Audio analysis to understand the characteristics of songs and match them with user preferences.
- NLP (Natural Language Processing) technology for processing user-generated content and feedback.
Need - Locating correct localities in South America to establish connections to the communication network and performing preventive maintenance for networks.
- Satellite data used to evaluate and locate localities.
- Frequency data for network planning and optimization.
- Network activity data for monitoring and maintaining network performance.
- Visual computing technology for analyzing satellite data and locating correct localities.
- Machine learning technology for network planning and optimization.
Need - Reducing the spread of fake news and spam from bots.
- Network activity data to monitor user interactions and content sharing.
- Online content data, including tweets and other user-generated content.
- Machine learning technology, particularly NLP (Natural Language Processing), for analyzing content and identifying fake news and spam.
- Improving service quality by considering external factors such as weather conditions and internal factors like usage levels and patterns.
- Implementing a chatbot for self-service customer support.
- Monitoring network activity to assess performance and identify areas for improvement.
- Environment and weather data to anticipate potential service disruptions.
- Network performance data for optimizing service delivery.
- Subscription renewal data to manage customer accounts and services.
- Machine learning technology for analyzing data and improving service quality.
- NLP (Natural Language Processing) for the chatbot to understand and respond to customer queries effectively.
Need - Improving video streaming performance and increasing viewership levels.
- Monitoring network activity to assess video streaming performance and identify areas for improvement.
- Viewing data to understand user preferences and trends.
- Social media data for analyzing audience engagement and feedback.
- Machine learning technology for analyzing data and optimizing video streaming performance.
- Machine learning algorithms for content recommendations to increase viewership levels.
Sector: Service, Financial, and Health
- Card fraud detection to enhance security and protect cardholders from fraudulent activities.
- Providing a virtual personal assistant for passengers to offer personalized services.
- Past purchase data to understand cardholders' spending patterns and preferences.
- Past fraud data for building models to detect and prevent fraudulent transactions.
- Machine learning, both supervised and unsupervised, for card fraud detection and improving security.
- Implementing virtual personal assistants using machine learning to offer personalized services to passengers.
- Providing support for medical decisions to assist healthcare professionals in making informed choices.
- Offering recommendations for medical, scientific research to enhance research outcomes.
- Contents of research journals to analyze and provide insights to healthcare professionals.
- Medical research books for comprehensive medical knowledge and information.
- Metapatient data for a holistic view of patient health across various aspects.
- Insurance claims data for analyzing medical procedures, treatments, and outcomes.
- NLP (Natural Language Processing) technology for analyzing and understanding medical literature and research papers.
- Utilizing NLP for extracting insights from medical records and meta patient data to aid medical decisions and research.
Need - Detecting product counterfeits online as a service company to prevent the spread of counterfeit products.
- Photographs of products to be used for counterfeit detection purposes.
- Additional customized data for enhancing accuracy and specificity.
- Advanced scanning technologies for capturing detailed product images and features.
- Machine learning for training the counterfeit detection models.
- Deep learning algorithms for more sophisticated counterfeit identification.
- Visual computing for analyzing and understanding product images.
Need - Simplifying, shortening, and potentially reducing the cost of mortgage approval processes.
- Request documents related to mortgage applications, including financial records and applicant information.
- Machine learning, particularly NLP (Natural Language Processing), for analyzing and processing mortgage application documents.
- Utilizing NLP to extract relevant information and streamline the mortgage approval processes.
Need - Increasing sales through accurate lead finding by leveraging Facebook services segmentation.
- Existing customer data, including information on consumption habits.
- Data on sites where customers spend time online.
- Response data to advertisements for understanding customer engagement.
- Machine learning, particularly NLP (Natural Language Processing), for analyzing customer data and identifying potential leads through Facebook services segmentation.
Need - Reducing the cost of booking flight tickets by learning the right timing for booking deals and providing precise destination and date recommendations to customers.
- Ticket order and price data (purchased for a fee) for understanding flight price patterns and trends.
- Consumer data for analyzing customer preferences and travel habits.
- Machine learning, particularly NLP (Natural Language Processing), for analyzing flight data and identifying optimal booking times and deals.
- Utilizing NLP to understand customer preferences and offer personalized flight recommendations.
Need - Early detection of cancer and stroke using advanced medical imaging technologies.
- Medical scans such as X-ray and CT scans for accurate analysis and detection of abnormalities.
- Medical information of patients for comprehensive assessment and diagnosis.
- Deep learning algorithms for analyzing medical scans and identifying potential signs of cancer and stroke.
- Visual computing for enhancing the accuracy of medical imaging and diagnostics.
Need - Reducing false claims to cancel a purchase charge by accurately identifying legitimate transactions.
- Purchase data to understand transaction details and patterns.
- Fraud data for identifying patterns of fraudulent activities.
- Machine learning, both supervised and unsupervised, for analyzing transaction data and detecting potential fraudulent claims.
- Utilizing machine learning algorithms to distinguish between legitimate transactions and false claims to ensure accurate charge processing.
- Helping organizations that use products better understand the characteristics of their customers through customer relationship management (CRM) services.
- Providing telephone service as part of their CRM service offerings.
- Customer data, including information on purchases, finances, and service interactions.
- Machine learning, particularly NLP (Natural Language Processing), for analyzing customer data and gaining insights into customer behavior and preferences.
- Improving all company activities, from sales to service experience, by speeding up the arrival of vehicles or food deliveries.
- Predicting if customers are drunk and preventing drivers who do not want to come.
- Predicting payment prospects for better financial management.
- Travel data, including trip details and routes taken.
- GPS data of passengers and drivers for real-time tracking and coordination.
- Map data for navigation and route optimization.
- Machine learning technology for optimizing transportation and food delivery services, reducing wait times, and improving customer experience.
- Utilizing machine learning algorithms for identifying drunk customers and preventing unwanted drivers.
- Predictive analytics using machine learning for predicting payment prospects and enhancing financial planning and management.
Sector: Heavy Industry, Automotive, Aerospace, Industry 4.0
- Developing autonomous vehicles to meet the growing demand for self-driving cars.
- Implementing driving adjustment features for passenger comfort and convenience.
- Driving data collected through cameras installed in cars for understanding road conditions and surroundings.
- GPS data for navigation and route planning.
- Cognitive computing for simulating human-like decision-making in autonomous driving systems.
- Machine learning algorithms for analyzing driving data and improving autonomous vehicle performance.
- Visual computing for interpreting visual information from cameras and making driving adjustments based on passenger preferences.
- Creating smart power generation stations for optimized operation and identifying potential inefficiencies.
- Efficient procurement management for better resource allocation.
- Data from sensors installed in power generation stations for monitoring operations and performance.
- Machine learning for analyzing data from power generation stations to optimize their operation and improve efficiency.
- Predictive analytics for identifying potential issues and inefficiencies in power generation processes.
- Utilizing machine learning algorithms for procurement management to optimize resource allocation and cost-effectiveness.
- Reducing pesticide contamination by recommending targeted use of substances to farmers.
- Providing recommendations to farmers on where to sow crops for optimized yields.
- Tumor photography data for identifying and managing pest infestations.
- Pest reservoir data to understand pest populations and distribution.
- Machine learning algorithms for analyzing data and recommending targeted pesticide use to minimize contamination.
- Utilizing machine learning to provide farmers with location-based recommendations on optimal crop planting.
- Improving and streamlining the entire energy supply chain, including balancing electricity loads, to enhance energy efficiency.
- Energy data collected from various facilities and equipment to monitor energy usage and performance.
- Machine learning for analyzing energy data and optimizing energy usage throughout the supply chain.
- Predictive analytics to forecast and manage electricity loads for better energy balancing and efficiency.
Need: Developing a Smart Rail System to achieve the following objectives:
- Improving train compliance with schedules and arrival times.
- Optimizing energy usage for efficient rail operations.
- Maximizing resource utilization for cost-effective rail services.
- Sensors installed in all rail system components for monitoring performance and maintenance needs.
- Cameras for detecting potential rail interferences and obstacles.
- Machine learning algorithms for analyzing data from sensors and cameras to optimize train schedules and improve compliance.
- Utilizing visual computing to process camera data and detect potential rail interferences.
Need: Developing autonomous vehicles and providing a virtual assistant for driver communication.
- Sensors in the car for real-time data on vehicle performance and surroundings.
- Shared data from all vehicle databases for insights and improvements based on collective information.
- Traffic cameras for monitoring road conditions and traffic patterns.
- Sensors on the way for collecting data on road infrastructure and potential obstacles.
- Passenger phone data for personalized in-car experiences.
- Machine learning algorithms for processing sensor data and enabling autonomous driving capabilities.
- Utilizing machine learning for analyzing shared data from vehicle databases to improve overall vehicle performance and safety.
- Implementing machine learning for the virtual assistant enhances driver communication and provides personalized assistance.
Need: The need to implement autonomous components for enhanced driving safety.
- Sensors in the car for real-time data collection.
- Sensors on the way to gather information about the road and surroundings.
- Machine learning for processing and analyzing data from sensors.
- Deep learning for developing advanced autonomous capabilities.
- Predictive analytics for anticipating potential risks and improving safety measures.
The book aims to provide an overview of AI capabilities while inspiring individuals to incorporate AI into their organizations and advance their career development. The main recommendations for readers within organizations are as follows:
1. Develop AI awareness.
2. Foster capabilities within the organization, not solely relying on consultants.
3. Ensure the appropriate handling of data.
4. Stay updated on the technological base, including cloud computing and intelligent equipment, and adapt your computing accordingly.
5. Use AI ethically.
6. Address and eliminate inherent biases.
7. Prepare yourself and advocate for restructuring your role for partial mechanization.
8. Consider exploring new professions such as machine learning engineering, natural language engineering, and more. The future is already here.